0xrsydn/idx-cli
1
mirror of https://github.com/0xrsydn/idx-cli.git synced 2026-11-07 09:45:36 +00:00
Code Issues Projects Releases 1 Packages Wiki Activity

Merge pull request #38 from 0xrsydn/test/agent-smoke

Browse source 
Agent smoke test + security and correctness fixes from Codex review
This commit is contained in:
brutaljokerz 2026-10-15 10:09:06 +07:00 • committed by GitHub
commit 9841a42173
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
23 changed files with 1408 additions and 107 deletions
Download patch file Download diff file Expand all files Collapse all files

2
AGENTS.md
View file

@ -22,6 +22,7 @@ crates/idx-cli/ binary `idx`
└── tests/ cli.rs (behavior), golden.rs + golden/ (output parity) └── tests/ cli.rs (behavior), golden.rs + golden/ (output parity)
crates/idx-mcp/ binary `idx-mcp`: stdio MCP tools (server.rs), inputs (params.rs) crates/idx-mcp/ binary `idx-mcp`: stdio MCP tools (server.rs), inputs (params.rs)
└── tests/mcp.rs JSON-RPC over stdio against the mock provider └── tests/mcp.rs JSON-RPC over stdio against the mock provider
tests/agents/ opt-in agent smoke test: real coding agents against idx-mcp
scripts/ smoke, installer, and ownership snapshot publisher scripts/ smoke, installer, and ownership snapshot publisher
``` ```
@ -52,4 +53,5 @@ Docs:
nix flake check -L # canonical gate (fmt, clippy, tests, package) nix flake check -L # canonical gate (fmt, clippy, tests, package)
nix develop --command cargo test <name> # focused loop nix develop --command cargo test <name> # focused loop
scripts/live-smoke.sh --compare <old> # live drift check, A/B against an older build scripts/live-smoke.sh --compare <old> # live drift check, A/B against an older build
tests/agents/agent-smoke.sh # after changing idx-mcp tools: real agents, opt-in
``` ```

2
Cargo.toml
View file

@ -5,7 +5,7 @@ members = ["crates/idx-core", "crates/idx-cli", "crates/idx-mcp"]
[workspace.package] [workspace.package]
version = "0.3.0" version = "0.3.0"
edition = "2024" edition = "2024"
rust-version = "1.85" rust-version = "1.89"
license = "MIT" license = "MIT"
repository = "https://github.com/0xrsydn/idx-cli" repository = "https://github.com/0xrsydn/idx-cli"
homepage = "https://github.com/0xrsydn/idx-cli" homepage = "https://github.com/0xrsydn/idx-cli"

4
README.md
View file

@ -47,7 +47,7 @@ helper tool below is available automatically.
cargo install idx-cli cargo install idx-cli
``` ```
Builds from source; requires Rust `1.85+`. Builds from source; requires Rust `1.89+`.
### Runtime helper tools ### Runtime helper tools
@ -172,7 +172,7 @@ client:
```bash ```bash
claude mcp add idx -- npx -y idx-mcp # Claude Code, via npm claude mcp add idx -- npx -y idx-mcp # Claude Code, via npm
cargo install idx-mcp # or build it (Rust 1.88+) cargo install idx-mcp # or build it (Rust 1.89+)
nix profile install github:0xrsydn/idx-cli#idx-mcp nix profile install github:0xrsydn/idx-cli#idx-mcp
``` ```

33
crates/idx-core/src/api/mod.rs
View file

@ -163,6 +163,19 @@ pub fn resolve_symbol(symbol: &str, exchange: &str) -> Result<String, IdxError>
"ticker symbol cannot be empty".into(), "ticker symbol cannot be empty".into(),
)); ));
} }
// Tickers name cache files and provider URLs: letters, digits, and the
// separators real symbols use (BBCA.JK, BRK-B, ^JKSE), nothing else.
if trimmed.len() > 32
|| trimmed.contains("..")
|| !trimmed
.chars()
.all(|ch| ch.is_ascii_alphanumeric() || matches!(ch, '.' | '-' | '^' | '='))
{
return Err(IdxError::InvalidInput(format!(
"invalid ticker symbol '{}'",
symbol.trim()
)));
}
if let Some((base, suffix)) = trimmed.rsplit_once('.') if let Some((base, suffix)) = trimmed.rsplit_once('.')
&& !base.is_empty() && !base.is_empty()
&& !suffix.is_empty() && !suffix.is_empty()
@ -301,6 +314,26 @@ mod tests {
assert_eq!(resolve_symbol("BBCA", "JK").unwrap(), "BBCA.JK"); assert_eq!(resolve_symbol("BBCA", "JK").unwrap(), "BBCA.JK");
} }
#[test]
fn rejects_symbols_that_are_not_tickers() {
for bad in [
"../../REPORT",
"a/b",
"a\\b",
"BB CA",
"BBCA..JK",
"BBCA\u{0}",
&"A".repeat(40),
] {
assert!(
matches!(resolve_symbol(bad, "JK"), Err(IdxError::InvalidInput(_))),
"{bad:?} accepted"
);
}
assert_eq!(resolve_symbol("^jkse", "JK").unwrap(), "^JKSE.JK");
assert_eq!(resolve_symbol("BRK-B.US", "JK").unwrap(), "BRK-B.US");
}
#[test] #[test]
fn selected_provider_exposes_expected_capabilities_for_msn() { fn selected_provider_exposes_expected_capabilities_for_msn() {
let provider = build_selected_provider(ProviderKind::Msn, false, false); let provider = build_selected_provider(ProviderKind::Msn, false, false);

25
crates/idx-core/src/api/yahoo/client.rs
View file

@ -2,6 +2,7 @@ use std::path::{Path, PathBuf};
use std::process::Command; use std::process::Command;
use std::process::Stdio; use std::process::Stdio;
use std::sync::Mutex; use std::sync::Mutex;
use std::sync::atomic::{AtomicU64, Ordering};
use std::time::Duration; use std::time::Duration;
use crate::api::types::{Interval, Period}; use crate::api::types::{Interval, Period};
@ -19,6 +20,9 @@ const CRUMB_FETCH_URL: &str = "https://query1.finance.yahoo.com/v1/test/getcrumb
pub(super) struct YahooClient { pub(super) struct YahooClient {
agent: ureq::Agent, agent: ureq::Agent,
crumb: Mutex<Option<String>>, crumb: Mutex<Option<String>>,
/// One jar per client, paired with its crumb: clients in the same process
/// (concurrent MCP calls) must not overwrite each other's cookies.
cookie_jar: PathBuf,
} }
impl YahooClient { impl YahooClient {
@ -29,9 +33,16 @@ impl YahooClient {
.build() .build()
.into(); .into();
static NEXT_JAR: AtomicU64 = AtomicU64::new(0);
let cookie_jar = std::env::temp_dir().join(format!(
"idx_yf_{}_{}.txt",
std::process::id(),
NEXT_JAR.fetch_add(1, Ordering::Relaxed)
));
Self { Self {
agent, agent,
crumb: Mutex::new(None), crumb: Mutex::new(None),
cookie_jar,
} }
} }
@ -49,10 +60,6 @@ impl YahooClient {
) )
} }
fn cookie_jar_path() -> PathBuf {
std::env::temp_dir().join(format!("idx_yf_{}.txt", std::process::id()))
}
pub(super) fn parse_crumb_body(raw: &str) -> Result<String, IdxError> { pub(super) fn parse_crumb_body(raw: &str) -> Result<String, IdxError> {
let crumb = raw.trim(); let crumb = raw.trim();
if crumb.is_empty() { if crumb.is_empty() {
@ -123,7 +130,7 @@ impl YahooClient {
fn fetch_crumb_via_curl(&self) -> Result<String, IdxError> { fn fetch_crumb_via_curl(&self) -> Result<String, IdxError> {
let binary = curl_impersonate::chrome_curl_binary()?; let binary = curl_impersonate::chrome_curl_binary()?;
let cookie_jar = Self::cookie_jar_path(); let cookie_jar = &self.cookie_jar;
let cookie_jar_str = cookie_jar.to_str().ok_or_else(|| { let cookie_jar_str = cookie_jar.to_str().ok_or_else(|| {
IdxError::Io(format!( IdxError::Io(format!(
"failed to encode Yahoo cookie jar path {}", "failed to encode Yahoo cookie jar path {}",
@ -235,7 +242,7 @@ impl YahooClient {
) -> Result<QuoteSummaryResponse, IdxError> { ) -> Result<QuoteSummaryResponse, IdxError> {
for auth_attempt in 0..2 { for auth_attempt in 0..2 {
let crumb = self.get_or_init_crumb()?; let crumb = self.get_or_init_crumb()?;
let cookie_header = match Self::cookie_header_from_jar(&Self::cookie_jar_path()) { let cookie_header = match Self::cookie_header_from_jar(&self.cookie_jar) {
Ok(header) => header, Ok(header) => header,
Err(err) => { Err(err) => {
runtime::warn(format!("failed to parse Yahoo cookie jar: {err}")); runtime::warn(format!("failed to parse Yahoo cookie jar: {err}"));
@ -284,6 +291,12 @@ fn jitter() -> Duration {
Duration::from_millis(fastrand::u64(0..100)) Duration::from_millis(fastrand::u64(0..100))
} }
impl Drop for YahooClient {
fn drop(&mut self) {
let _ = std::fs::remove_file(&self.cookie_jar);
}
}
#[cfg(test)] #[cfg(test)]
mod tests { mod tests {
use super::YahooClient; use super::YahooClient;

70
crates/idx-core/src/cache.rs
View file

@ -136,7 +136,14 @@ impl Cache {
}; };
let raw = serde_json::to_string_pretty(&entry) let raw = serde_json::to_string_pretty(&entry)
.map_err(|e| IdxError::ParseError(e.to_string()))?; .map_err(|e| IdxError::ParseError(e.to_string()))?;
fs::write(path, raw).map_err(|e| IdxError::Io(e.to_string())) // Write then rename, so a concurrent reader never sees a partial
// entry (and deletes it as corrupt).
let temp = path.with_extension(format!("json.{}.tmp", fastrand::u64(..)));
fs::write(&temp, raw).map_err(|e| IdxError::Io(e.to_string()))?;
fs::rename(&temp, &path).map_err(|e| {
let _ = fs::remove_file(&temp);
IdxError::Io(e.to_string())
})
} }
pub fn info(&self) -> Result<CacheInfo, IdxError> { pub fn info(&self) -> Result<CacheInfo, IdxError> {
@ -226,7 +233,8 @@ impl Cache {
runtime::warn(format!( runtime::warn(format!(
"corrupted cache entry for {data_type}/{symbol}, treating as miss: {e}" "corrupted cache entry for {data_type}/{symbol}, treating as miss: {e}"
)); ));
let _ = fs::remove_file(&path); // Not deleted: a concurrent writer may have just replaced it
// with a valid entry. The next put overwrites it atomically.
return Ok(None); return Ok(None);
} }
}; };
@ -237,17 +245,36 @@ impl Cache {
entry.schema_version, entry.schema_version,
CURRENT_SCHEMA_VERSION CURRENT_SCHEMA_VERSION
)); ));
let _ = fs::remove_file(&path);
return Ok(None); return Ok(None);
} }
Ok(Some(entry)) Ok(Some(entry))
} }
fn entry_path(&self, data_type: &str, symbol: &str) -> PathBuf { fn entry_path(&self, data_type: &str, symbol: &str) -> PathBuf {
self.root.join(data_type).join(format!("{symbol}.json")) self.root
.join(safe_file_name(data_type))
.join(format!("{}.json", safe_file_name(symbol)))
} }
} }
/// Keys become single path components: anything that could separate or
/// climb directories is percent-encoded. Keys made of letters, digits, and
/// `._-:^=` (every key idx produces) are unchanged.
fn safe_file_name(key: &str) -> String {
let mut out = String::with_capacity(key.len());
for (i, byte) in key.bytes().enumerate() {
let safe = byte.is_ascii_alphanumeric()
|| matches!(byte, b'_' | b'-' | b':' | b'^' | b'=')
|| (byte == b'.' && i > 0);
if safe {
out.push(byte as char);
} else {
out.push_str(&format!("%{byte:02X}"));
}
}
out
}
fn is_expired(fetched_at: DateTime<Utc>, ttl_secs: u64) -> bool { fn is_expired(fetched_at: DateTime<Utc>, ttl_secs: u64) -> bool {
let ttl = let ttl =
chrono::Duration::from_std(Duration::from_secs(ttl_secs)).unwrap_or(chrono::Duration::MAX); chrono::Duration::from_std(Duration::from_secs(ttl_secs)).unwrap_or(chrono::Duration::MAX);
@ -275,7 +302,7 @@ mod tests {
use serde::{Deserialize, Serialize}; use serde::{Deserialize, Serialize};
use super::{Cache, CacheEntry}; use super::{Cache, CacheEntry, safe_file_name};
static TMP_COUNTER: AtomicUsize = AtomicUsize::new(0); static TMP_COUNTER: AtomicUsize = AtomicUsize::new(0);
@ -325,7 +352,7 @@ mod tests {
} }
#[test] #[test]
fn corrupted_cache_entry_returns_none_and_deletes_file() { fn corrupted_cache_entry_is_a_miss_until_replaced() {
let root = tmp(); let root = tmp();
let cache = Cache::at(root.clone()); let cache = Cache::at(root.clone());
@ -344,12 +371,16 @@ mod tests {
.expect("get should not error"); .expect("get should not error");
assert_eq!(result, None, "corrupted entry should be treated as miss"); assert_eq!(result, None, "corrupted entry should be treated as miss");
// The corrupted file should be deleted // Left for the next put to replace (deleting could race a writer).
assert!(!path.exists(), "corrupted file should be deleted"); cache
.put("quote", "CORRUPT.JK", &T { v: 7 }, 3600)
.expect("put");
let result: Option<T> = cache.get("quote", "CORRUPT.JK").expect("get");
assert_eq!(result, Some(T { v: 7 }));
} }
#[test] #[test]
fn corrupted_cache_entry_get_stale_returns_none_and_deletes_file() { fn corrupted_cache_entry_get_stale_is_a_miss() {
let root = tmp(); let root = tmp();
let cache = Cache::at(root.clone()); let cache = Cache::at(root.clone());
@ -374,7 +405,24 @@ mod tests {
"corrupted entry should be treated as miss in get_stale" "corrupted entry should be treated as miss in get_stale"
); );
// The corrupted file should be deleted assert!(path.exists(), "left for the next put to replace");
assert!(!path.exists(), "corrupted file should be deleted"); }
#[test]
fn cache_keys_cannot_leave_the_cache_root() {
assert_eq!(safe_file_name("BBCA.JK-3mo-1d"), "BBCA.JK-3mo-1d");
assert_eq!(
safe_file_name("top-performers:id:50"),
"top-performers:id:50"
);
assert_eq!(safe_file_name("../../REPORT"), "%2E.%2F..%2FREPORT");
assert_eq!(safe_file_name(".."), "%2E.");
let cache = Cache::at(std::path::PathBuf::from("/cache/idx"));
let path = cache.entry_path("msn-quote", "../../REPORT");
assert_eq!(
path.parent().unwrap(),
std::path::Path::new("/cache/idx/msn-quote")
);
} }
} }

118
crates/idx-core/src/investor.rs
View file

@ -47,18 +47,49 @@ impl Investor {
} }
} }
fn connect(&self) -> Result<Connection, IdxError> { /// Cross-process lock beside the database. Writers (sync install, import,
/// entity edits) hold it exclusively; readers hold it shared, so a
/// snapshot swap never happens under an open connection. Released when
/// the returned file is dropped.
fn lock_file(&self, exclusive: bool) -> Result<fs::File, IdxError> {
// Every path to one database (symlinks included) shares one lock.
let real = db::resolve_db_path(&self.db_path()?)?;
lock_at(&real, exclusive)
}
fn write_lock(&self) -> Result<fs::File, IdxError> {
self.lock_file(true)
}
/// A read connection holding the shared lock for its lifetime.
fn connect(&self) -> Result<ReadConnection, IdxError> {
let real = db::resolve_db_path(&self.db_path()?)?;
let lock = match lock_at(&real, false) {
Ok(lock) => Some(lock),
// No lock file and none can be created (a read-only filesystem):
// no writer can lock it to replace the database either.
Err(_) if unlockable(&real) => None,
Err(err) => return Err(err),
};
Ok(ReadConnection {
conn: self.open_db()?,
_lock: lock,
})
}
/// For writers, which already hold the exclusive lock.
fn open_db(&self) -> Result<Connection, IdxError> {
db::open_db_at(&self.db_path()?) db::open_db_at(&self.db_path()?)
} }
pub fn releases(&self) -> Result<Vec<OwnershipRelease>, IdxError> { pub fn releases(&self) -> Result<Vec<OwnershipRelease>, IdxError> {
db::query_releases(&self.connect()?) db::query_releases(&*self.connect()?)
} }
/// Whether the database has data, and which releases it holds. /// Whether the database has data, and which releases it holds.
pub fn status(&self) -> Result<OwnershipStatus, IdxError> { pub fn status(&self) -> Result<OwnershipStatus, IdxError> {
let db_path = self.db_path()?; let db_path = self.db_path()?;
let releases = db::query_releases(&db::open_db_at(&db_path)?)?; let releases = db::query_releases(&*self.connect()?)?;
Ok(OwnershipStatus { Ok(OwnershipStatus {
db_path: db_path.display().to_string(), db_path: db_path.display().to_string(),
populated: !releases.is_empty(), populated: !releases.is_empty(),
@ -116,7 +147,7 @@ impl Investor {
} }
pub fn search(&self, query: &str, limit: usize) -> Result<Vec<Entity>, IdxError> { pub fn search(&self, query: &str, limit: usize) -> Result<Vec<Entity>, IdxError> {
search::fts_search(&self.connect()?, query, limit) search::fts_search(&*self.connect()?, query, limit)
} }
pub fn cross_holders( pub fn cross_holders(
@ -124,7 +155,7 @@ impl Investor {
min_tickers: usize, min_tickers: usize,
top: usize, top: usize,
) -> Result<Vec<CrossHolderRow>, IdxError> { ) -> Result<Vec<CrossHolderRow>, IdxError> {
db::query_cross_holders(&self.connect()?, min_tickers, top) db::query_cross_holders(&*self.connect()?, min_tickers, top)
} }
pub fn concentration( pub fn concentration(
@ -132,7 +163,7 @@ impl Investor {
by: ConcentrationRank, by: ConcentrationRank,
top: usize, top: usize,
) -> Result<Vec<(String, ConcentrationMetrics)>, IdxError> { ) -> Result<Vec<(String, ConcentrationMetrics)>, IdxError> {
db::query_concentration(&self.connect()?, by.as_str(), top) db::query_concentration(&*self.connect()?, by.as_str(), top)
} }
/// Institutional (Bing) flow; `None` when no flow data exists. /// Institutional (Bing) flow; `None` when no flow data exists.
@ -151,7 +182,7 @@ impl Investor {
to: &str, to: &str,
ticker: Option<&str>, ticker: Option<&str>,
) -> Result<Vec<ChangeRow>, IdxError> { ) -> Result<Vec<ChangeRow>, IdxError> {
let mut rows = db::query_changes(&self.connect()?, from, to)?; let mut rows = db::query_changes(&*self.connect()?, from, to)?;
if let Some(ticker) = ticker { if let Some(ticker) = ticker {
let ticker = ticker.trim().to_uppercase(); let ticker = ticker.trim().to_uppercase();
rows.retain(|row| row.ticker_code == ticker); rows.retain(|row| row.ticker_code == ticker);
@ -164,20 +195,22 @@ impl Investor {
root: &str, root: &str,
depth: usize, depth: usize,
) -> Result<(Vec<GraphNode>, Vec<GraphEdge>), IdxError> { ) -> Result<(Vec<GraphNode>, Vec<GraphEdge>), IdxError> {
graph::query_ownership_graph(&self.connect()?, root, depth) graph::query_ownership_graph(&*self.connect()?, root, depth)
} }
pub fn unresolved(&self, limit: usize) -> Result<Vec<UnresolvedRow>, IdxError> { pub fn unresolved(&self, limit: usize) -> Result<Vec<UnresolvedRow>, IdxError> {
db::list_unresolved(&self.connect()?, limit) db::list_unresolved(&*self.connect()?, limit)
} }
pub fn map_alias(&self, alias: &str, entity: &str) -> Result<(), IdxError> { pub fn map_alias(&self, alias: &str, entity: &str) -> Result<(), IdxError> {
db::manual_map(&self.connect()?, alias, entity) let _lock = self.write_lock()?;
db::manual_map(&self.open_db()?, alias, entity)
} }
/// Merges entity `merge` into `keep`, reporting how many rows moved. /// Merges entity `merge` into `keep`, reporting how many rows moved.
pub fn merge_entities(&self, keep: i64, merge: i64) -> Result<MergeResult, IdxError> { pub fn merge_entities(&self, keep: i64, merge: i64) -> Result<MergeResult, IdxError> {
let conn = self.connect()?; let _lock = self.write_lock()?;
let conn = self.open_db()?;
let count = |table: &str| -> Result<i64, IdxError> { let count = |table: &str| -> Result<i64, IdxError> {
conn.query_row( conn.query_row(
&format!("SELECT COUNT(*) FROM {table} WHERE entity_id = ?1"), &format!("SELECT COUNT(*) FROM {table} WHERE entity_id = ?1"),
@ -216,15 +249,17 @@ impl Investor {
force: bool, force: bool,
) -> Result<OwnershipSyncResult, IdxError> { ) -> Result<OwnershipSyncResult, IdxError> {
let manifest_source = snapshot::resolve_manifest_source(manifest)?; let manifest_source = snapshot::resolve_manifest_source(manifest)?;
let db_path = self.db_path()?; snapshot::sync_snapshot(&manifest_source, &self.db_path()?, force, lock_at)
snapshot::sync_snapshot(&manifest_source, &db_path, force)
} }
/// Imports one report. A release already imported (same SHA-256) is /// Imports one report. A release already imported (same SHA-256) is
/// skipped unless `force`, which replaces it. /// skipped unless `force`, which replaces it.
pub fn import(&self, source: &ImportSource, force: bool) -> Result<ImportResult, IdxError> { pub fn import(&self, source: &ImportSource, force: bool) -> Result<ImportResult, IdxError> {
// Validate and download first (unchanged error order), then lock
// only around the database writes.
let input = self.resolve_import_input(source)?; let input = self.resolve_import_input(source)?;
let conn = self.connect()?; let _lock = self.write_lock()?;
let conn = self.open_db()?;
let sha256 = sha256_file(&input.path)?; let sha256 = sha256_file(&input.path)?;
if !force && db::release_exists(&conn, &sha256)? { if !force && db::release_exists(&conn, &sha256)? {
@ -412,6 +447,61 @@ pub struct OwnershipStatus {
pub releases: Vec<OwnershipRelease>, pub releases: Vec<OwnershipRelease>,
} }
fn lock_path(db_path: &Path) -> PathBuf {
let mut name = db_path.as_os_str().to_os_string();
name.push(".lock");
PathBuf::from(name)
}
/// Locks `<db>.lock` beside the resolved database path. A shared lock does
/// not need write access, so readers fall back to opening an existing lock
/// file read-only (e.g. one created by another user).
fn lock_at(db_path: &Path, exclusive: bool) -> Result<fs::File, IdxError> {
let path = lock_path(db_path);
let opened = fs::OpenOptions::new()
.create(true)
.truncate(false)
.write(true)
.open(&path);
let file = match opened {
Ok(file) => file,
Err(_) if !exclusive && path.exists() => fs::File::open(&path)
.map_err(|e| IdxError::Io(format!("failed to open ownership lock: {e}")))?,
Err(e) => return Err(IdxError::Io(format!("failed to open ownership lock: {e}"))),
};
if exclusive {
file.lock()
} else {
file.lock_shared()
}
.map_err(|e| IdxError::Io(format!("failed to lock ownership database: {e}")))?;
Ok(file)
}
/// True when the lock file is absent and cannot be created.
fn unlockable(db_path: &Path) -> bool {
let path = lock_path(db_path);
!path.exists()
&& matches!(
fs::OpenOptions::new().create(true).truncate(false).write(true).open(&path),
Err(e) if matches!(e.kind(), std::io::ErrorKind::PermissionDenied | std::io::ErrorKind::ReadOnlyFilesystem)
)
}
struct ReadConnection {
// Dropped (closed) before the lock it is declared ahead of.
conn: Connection,
_lock: Option<fs::File>,
}
impl std::ops::Deref for ReadConnection {
type Target = Connection;
fn deref(&self) -> &Connection {
&self.conn
}
}
/// Holder sources for `Investor::ticker`. /// Holder sources for `Investor::ticker`.
#[derive(Debug, Clone, Copy, PartialEq, Eq)] #[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum HolderSource { pub enum HolderSource {

21
crates/idx-core/src/ownership/archive.rs
View file

@ -1,5 +1,5 @@
use std::fs; use std::fs;
use std::io::{Cursor, Read}; use std::io::Cursor;
use std::path::Path; use std::path::Path;
use chrono::NaiveDate; use chrono::NaiveDate;
@ -177,11 +177,7 @@ fn extract_txt_from_zip(path: &Path) -> Result<String, IdxError> {
continue; continue;
} }
let mut output = String::new(); return super::read_zip_entry_bounded(&mut file, "KSEI archive TXT entry");
file.read_to_string(&mut output).map_err(|e| {
IdxError::ParseError(format!("failed to decode KSEI archive TXT entry: {e}"))
})?;
return Ok(output);
} }
Err(IdxError::ParseError( Err(IdxError::ParseError(
@ -201,7 +197,8 @@ fn validate_header(header: &str) -> Result<(), IdxError> {
fn parse_archive_date(raw: &str) -> Result<NaiveDate, IdxError> { fn parse_archive_date(raw: &str) -> Result<NaiveDate, IdxError> {
let trimmed = raw.trim(); let trimmed = raw.trim();
if trimmed.len() != 11 { // Fixed byte offsets below; non-ASCII input would split a character.
if trimmed.len() != 11 || !trimmed.is_ascii() {
return Err(IdxError::ParseError(format!( return Err(IdxError::ParseError(format!(
"invalid KSEI archive date `{trimmed}`" "invalid KSEI archive date `{trimmed}`"
))); )));
@ -357,4 +354,14 @@ mod tests {
assert!(pdf_local_cp.total_shares <= archive_local_cp.total_shares); assert!(pdf_local_cp.total_shares <= archive_local_cp.total_shares);
assert!(pdf_local_cp.percentage_bps <= archive_local_cp.percentage_bps); assert!(pdf_local_cp.percentage_bps <= archive_local_cp.percentage_bps);
} }
#[test]
fn non_ascii_archive_dates_are_parse_errors_not_panics() {
// 11 bytes, but byte offsets would split the 'é'.
assert!(super::parse_archive_date("1éJan-2026").is_err());
assert_eq!(
super::parse_archive_date("27-FEB-2026").unwrap(),
chrono::NaiveDate::from_ymd_opt(2026, 2, 27).unwrap()
);
}
} }

75
crates/idx-core/src/ownership/db.rs
View file

@ -133,6 +133,43 @@ pub fn open_db() -> Result<Connection, IdxError> {
} }
/// Open the ownership database at `db_path`, creating it and migrating the schema. /// Open the ownership database at `db_path`, creating it and migrating the schema.
/// Where the database file really lives: symlinks are followed by reading
/// the link, so this works even while the target does not exist (a fresh
/// install, or mid-replacement during sync). Creates the parent directory.
pub(crate) fn resolve_db_path(path: &Path) -> Result<PathBuf, IdxError> {
let mut current = path.to_path_buf();
for _ in 0..16 {
let parent = match current.parent() {
Some(parent) if !parent.as_os_str().is_empty() => parent.to_path_buf(),
_ => PathBuf::from("."),
};
fs::create_dir_all(&parent).map_err(|e| IdxError::Io(e.to_string()))?;
let parent = fs::canonicalize(&parent).map_err(|e| IdxError::Io(e.to_string()))?;
let name = current.file_name().ok_or_else(|| {
IdxError::ConfigError(format!(
"ownership database path has no file name: {}",
path.display()
))
})?;
let candidate = parent.join(name);
match fs::symlink_metadata(&candidate) {
Ok(meta) if meta.file_type().is_symlink() => {
let target = fs::read_link(&candidate).map_err(|e| IdxError::Io(e.to_string()))?;
current = if target.is_absolute() {
target
} else {
parent.join(target)
};
}
_ => return Ok(candidate),
}
}
Err(IdxError::Io(format!(
"too many symlink levels resolving {}",
path.display()
)))
}
pub fn open_db_at(db_path: &Path) -> Result<Connection, IdxError> { pub fn open_db_at(db_path: &Path) -> Result<Connection, IdxError> {
if let Some(parent) = db_path.parent() { if let Some(parent) = db_path.parent() {
fs::create_dir_all(parent).map_err(|e| IdxError::DatabaseError(e.to_string()))?; fs::create_dir_all(parent).map_err(|e| IdxError::DatabaseError(e.to_string()))?;
@ -1196,7 +1233,10 @@ pub fn compute_concentration(percentages_bps: &[i64]) -> ConcentrationMetrics {
let top1_bps = values.first().copied().unwrap_or(0); let top1_bps = values.first().copied().unwrap_or(0);
let top3_bps: i64 = values.iter().take(3).sum(); let top3_bps: i64 = values.iter().take(3).sum();
let total_bps: i64 = values.iter().sum(); let total_bps: i64 = values.iter().sum();
let hhi: i64 = values.iter().map(|p| (p * p) / 10000).sum(); let hhi: i64 = values
.iter()
.map(|p| p.saturating_mul(*p) / 10000)
.fold(0, i64::saturating_add);
ConcentrationMetrics { ConcentrationMetrics {
top1_bps, top1_bps,
@ -1962,4 +2002,37 @@ mod tests {
.unwrap(); .unwrap();
assert_eq!(buyers, 1); assert_eq!(buyers, 1);
} }
#[cfg(unix)]
#[test]
fn resolve_db_path_follows_symlinks_even_when_the_target_is_missing() {
let root = std::env::temp_dir().join(format!(
"idx-resolve-{}-{}",
std::process::id(),
fastrand::u64(..)
));
let real_dir = root.join("real");
std::fs::create_dir_all(&real_dir).unwrap();
let real_dir = std::fs::canonicalize(&real_dir).unwrap();
let link = root.join("link.db");
// Dangling: the target does not exist yet (fresh install, mid-swap).
std::os::unix::fs::symlink(real_dir.join("ownership.db"), &link).unwrap();
assert_eq!(
super::resolve_db_path(&link).unwrap(),
real_dir.join("ownership.db")
);
// Relative link, resolved against the link's directory.
let relative = root.join("relative.db");
std::os::unix::fs::symlink("real/ownership.db", &relative).unwrap();
assert_eq!(
super::resolve_db_path(&relative).unwrap(),
real_dir.join("ownership.db")
);
// A plain path is returned as-is (canonical parent).
assert_eq!(
super::resolve_db_path(&real_dir.join("ownership.db")).unwrap(),
real_dir.join("ownership.db")
);
let _ = std::fs::remove_dir_all(&root);
}
} }

14
crates/idx-core/src/ownership/entities.rs
View file

@ -81,7 +81,11 @@ pub fn parse_id_percentage(s: &str) -> Result<i64, IdxError> {
IdxError::ParseError(format!("failed to parse Indonesian percentage '{s}': {e}")) IdxError::ParseError(format!("failed to parse Indonesian percentage '{s}': {e}"))
})?; })?;
Ok((whole_i * 100) + frac_i) whole_i
.checked_mul(100)
.and_then(|bps| bps.checked_add(frac_i))
.filter(|bps| (0..=10_000).contains(bps))
.ok_or_else(|| IdxError::ParseError(format!("Indonesian percentage '{s}' is out of range")))
} }
/// Parse KSEI date format to NaiveDate. /// Parse KSEI date format to NaiveDate.
@ -453,4 +457,12 @@ mod tests {
assert!(normalize_ksei_row(&raw).is_err()); assert!(normalize_ksei_row(&raw).is_err());
} }
#[test]
fn huge_percentages_are_parse_errors_not_overflow() {
assert!(super::parse_id_percentage("9223372036854775807,99").is_err());
assert!(super::parse_id_percentage("40000000,00").is_err());
assert_eq!(super::parse_id_percentage("100,00").unwrap(), 10_000);
assert_eq!(super::parse_id_percentage("54,94").unwrap(), 5494);
}
} }

24
crates/idx-core/src/ownership/mod.rs
View file

@ -8,3 +8,27 @@ pub mod search;
pub mod snapshot; pub mod snapshot;
pub mod types; pub mod types;
pub mod xlsx; pub mod xlsx;
use std::io::Read;
use crate::error::IdxError;
/// Largest decompressed ZIP entry read into memory. Real IDX worksheets and
/// KSEI TXT archives are tens of MB; this bounds a hostile "zip bomb".
const MAX_ZIP_ENTRY_BYTES: u64 = 256 * 1024 * 1024;
/// Reads one decompressed ZIP entry as UTF-8, failing past the size limit.
pub(crate) fn read_zip_entry_bounded(entry: impl Read, what: &str) -> Result<String, IdxError> {
let mut bytes = Vec::new();
entry
.take(MAX_ZIP_ENTRY_BYTES + 1)
.read_to_end(&mut bytes)
.map_err(|e| IdxError::ParseError(format!("failed to read {what}: {e}")))?;
if bytes.len() as u64 > MAX_ZIP_ENTRY_BYTES {
return Err(IdxError::ParseError(format!(
"{what} expands beyond {MAX_ZIP_ENTRY_BYTES} bytes"
)));
}
String::from_utf8(bytes)
.map_err(|e| IdxError::ParseError(format!("failed to decode {what}: {e}")))
}

128
crates/idx-core/src/ownership/snapshot.rs
View file

@ -144,20 +144,34 @@ pub fn parse_manifest(raw: &str) -> Result<OwnershipSnapshotManifest, IdxError>
Ok(manifest) Ok(manifest)
} }
pub fn sync_snapshot( /// `lock(resolved_db_path, exclusive)` takes the database's lock: shared
/// while inspecting the local database, exclusive while rechecking and
/// installing, and never held during the downloads. The path is resolved
/// once, so the lock and the replaced file are always the same database.
pub fn sync_snapshot<L>(
manifest_source: &str, manifest_source: &str,
db_path: &Path, db_path: &Path,
force: bool, force: bool,
lock: impl Fn(&Path, bool) -> Result<L, IdxError>,
) -> Result<OwnershipSyncResult, IdxError> { ) -> Result<OwnershipSyncResult, IdxError> {
// Results report the configured path; files are read and replaced at the
// resolved one, so a symlinked path updates its target instead of being
// overwritten with a regular file.
let configured_db_path = db_path;
let resolved_db_path = db::resolve_db_path(db_path)?;
let db_path = resolved_db_path.as_path();
let manifest = fetch_manifest(manifest_source)?; let manifest = fetch_manifest(manifest_source)?;
let local_state = inspect_local_db(db_path)?; let local_state = {
let _shared = lock(db_path, false)?;
inspect_local_db(db_path)?
};
let decision = build_sync_decision(local_state.as_ref(), &manifest.snapshot, force); let decision = build_sync_decision(local_state.as_ref(), &manifest.snapshot, force);
if !decision.should_download { if !decision.should_download {
return Ok(OwnershipSyncResult { return Ok(OwnershipSyncResult {
action: decision.action, action: decision.action,
manifest: manifest_source.to_string(), manifest: manifest_source.to_string(),
db_path: db_path.display().to_string(), db_path: configured_db_path.display().to_string(),
snapshot_version: manifest.snapshot.version.clone(), snapshot_version: manifest.snapshot.version.clone(),
latest_as_of_date: manifest.snapshot.latest_as_of_date, latest_as_of_date: manifest.snapshot.latest_as_of_date,
release_count: manifest.snapshot.release_count, release_count: manifest.snapshot.release_count,
@ -202,18 +216,28 @@ pub fn sync_snapshot(
let install_result = (|| -> Result<(), IdxError> { let install_result = (|| -> Result<(), IdxError> {
validate_snapshot_db(&temp_path, &manifest.snapshot)?; validate_snapshot_db(&temp_path, &manifest.snapshot)?;
let _lock = lock(db_path, true)?;
// The decision was made before the download; an import or edit that
// landed since must not be overwritten without --force.
if !force && !same_local_state(local_state.as_ref(), inspect_local_db(db_path)?.as_ref()) {
return Err(IdxError::DatabaseError(
"the local ownership database changed during sync; rerun `idx ownership sync`"
.to_string(),
));
}
install_snapshot_file(&temp_path, db_path) install_snapshot_file(&temp_path, db_path)
})(); })();
if install_result.is_err() { if install_result.is_err() {
let _ = fs::remove_file(&temp_path); let _ = fs::remove_file(&temp_path);
let _ = remove_sqlite_sidecars(&temp_path);
} }
install_result?; install_result?;
Ok(OwnershipSyncResult { Ok(OwnershipSyncResult {
action: decision.action, action: decision.action,
manifest: manifest_source.to_string(), manifest: manifest_source.to_string(),
db_path: db_path.display().to_string(), db_path: configured_db_path.display().to_string(),
snapshot_version: manifest.snapshot.version.clone(), snapshot_version: manifest.snapshot.version.clone(),
latest_as_of_date: manifest.snapshot.latest_as_of_date, latest_as_of_date: manifest.snapshot.latest_as_of_date,
release_count: manifest.snapshot.release_count, release_count: manifest.snapshot.release_count,
@ -340,8 +364,18 @@ fn inspect_local_db(db_path: &Path) -> Result<Option<LocalSnapshotState>, IdxErr
return Ok(None); return Ok(None);
} }
// Schema setup first: it may write to an empty database, and hashing
// afterwards keeps two inspections of an unchanged database equal. A
// published snapshot already has the schema, so its bytes are untouched.
let releases = {
let conn = Connection::open(db_path).map_err(|e| {
IdxError::DatabaseError(format!("failed to open {}: {e}", db_path.display()))
})?;
db::ensure_schema(&conn)?;
db::query_releases(&conn)?
};
let file_sha256 = if checkpoint_wal(db_path) { let file_sha256 = if checkpoint_wal(db_path) {
// Hash before opening: schema setup may write to an empty database.
let bytes = fs::read(db_path) let bytes = fs::read(db_path)
.map_err(|e| IdxError::Io(format!("failed to read {}: {e}", db_path.display())))?; .map_err(|e| IdxError::Io(format!("failed to read {}: {e}", db_path.display())))?;
Some(sha256_hex(&bytes)) Some(sha256_hex(&bytes))
@ -349,12 +383,6 @@ fn inspect_local_db(db_path: &Path) -> Result<Option<LocalSnapshotState>, IdxErr
None None
}; };
let conn = Connection::open(db_path).map_err(|e| {
IdxError::DatabaseError(format!("failed to open {}: {e}", db_path.display()))
})?;
db::ensure_schema(&conn)?;
let releases = db::query_releases(&conn)?;
Ok(Some(LocalSnapshotState { Ok(Some(LocalSnapshotState {
file_sha256, file_sha256,
latest_release: releases.first().cloned(), latest_release: releases.first().cloned(),
@ -362,6 +390,21 @@ fn inspect_local_db(db_path: &Path) -> Result<Option<LocalSnapshotState>, IdxErr
})) }))
} }
fn same_local_state(
before: Option<&LocalSnapshotState>,
after: Option<&LocalSnapshotState>,
) -> bool {
match (before, after) {
(None, None) => true,
(Some(before), Some(after)) => {
before.file_sha256.is_some()
&& before.file_sha256 == after.file_sha256
&& before.release_count == after.release_count
}
_ => false,
}
}
/// Moves committed WAL frames into the database file, so its hash covers local /// Moves committed WAL frames into the database file, so its hash covers local
/// edits too. Returns false when changes remain in the WAL. /// edits too. Returns false when changes remain in the WAL.
fn checkpoint_wal(db_path: &Path) -> bool { fn checkpoint_wal(db_path: &Path) -> bool {
@ -446,10 +489,12 @@ fn build_sync_decision(
}; };
} }
if snapshot // A rebuild never silently drops history the local database has.
.replaces_sqlite_sha256 if snapshot.release_count >= local.release_count
.iter() && snapshot
.any(|sha256| file_sha256.is_some_and(|local| sha256.eq_ignore_ascii_case(local))) .replaces_sqlite_sha256
.iter()
.any(|sha256| file_sha256.is_some_and(|local| sha256.eq_ignore_ascii_case(local)))
{ {
return SyncDecision { return SyncDecision {
action: OwnershipSyncAction::Updated, action: OwnershipSyncAction::Updated,
@ -462,7 +507,9 @@ fn build_sync_decision(
} }
if latest.sha256 == snapshot.latest_release_sha256 { if latest.sha256 == snapshot.latest_release_sha256 {
if latest.row_count != snapshot.latest_row_count { if latest.row_count != snapshot.latest_row_count
&& snapshot.release_count >= local.release_count
{
return SyncDecision { return SyncDecision {
action: OwnershipSyncAction::Updated, action: OwnershipSyncAction::Updated,
should_download: true, should_download: true,
@ -608,6 +655,8 @@ fn build_temp_path(db_path: &Path) -> PathBuf {
} }
fn install_snapshot_file(temp_path: &Path, db_path: &Path) -> Result<(), IdxError> { fn install_snapshot_file(temp_path: &Path, db_path: &Path) -> Result<(), IdxError> {
// Validation opened the temp database; its sidecars are named after it.
remove_sqlite_sidecars(temp_path)?;
if let Some(parent) = db_path.parent() if let Some(parent) = db_path.parent()
&& !parent.as_os_str().is_empty() && !parent.as_os_str().is_empty()
{ {
@ -717,7 +766,8 @@ fn sha256_hex(bytes: &[u8]) -> String {
fn read_text(source: &str, context: &str) -> Result<String, IdxError> { fn read_text(source: &str, context: &str) -> Result<String, IdxError> {
if is_http_source(source) { if is_http_source(source) {
let response = ureq::get(source) let response = http_agent()
.get(source)
.header("User-Agent", USER_AGENT) .header("User-Agent", USER_AGENT)
.header("Accept", "application/json,text/plain;q=0.9,*/*;q=0.8") .header("Accept", "application/json,text/plain;q=0.9,*/*;q=0.8")
.call() .call()
@ -733,9 +783,20 @@ fn read_text(source: &str, context: &str) -> Result<String, IdxError> {
.map_err(|e| IdxError::Io(format!("failed to read {context} {}: {e}", path.display()))) .map_err(|e| IdxError::Io(format!("failed to read {context} {}: {e}", path.display())))
} }
/// Downloads give up rather than hang: a stalled connection must not keep
/// sync (and anything waiting on it) blocked forever.
fn http_agent() -> ureq::Agent {
ureq::Agent::config_builder()
.timeout_connect(Some(std::time::Duration::from_secs(15)))
.timeout_global(Some(std::time::Duration::from_secs(600)))
.build()
.into()
}
fn read_bytes(source: &str, context: &str, limit_bytes: u64) -> Result<Vec<u8>, IdxError> { fn read_bytes(source: &str, context: &str, limit_bytes: u64) -> Result<Vec<u8>, IdxError> {
if is_http_source(source) { if is_http_source(source) {
let response = ureq::get(source) let response = http_agent()
.get(source)
.header("User-Agent", USER_AGENT) .header("User-Agent", USER_AGENT)
.header( .header(
"Accept", "Accept",
@ -909,6 +970,20 @@ mod tests {
assert!(!decision.should_download); assert!(!decision.should_download);
} }
#[test]
fn build_sync_decision_never_drops_history_through_a_rebuild() {
// A forced same-date rebuild with fewer releases lists the old build;
// a client holding it keeps its history instead of downgrading.
let earlier = "e".repeat(64);
let mut snapshot = same_date_snapshot(vec![earlier.clone()]);
snapshot.release_count = 1;
let mut local = same_release_local(&snapshot, &earlier);
local.release_count = 3;
let decision = build_sync_decision(Some(&local), &snapshot, false);
assert!(!decision.should_download, "{}", decision.reason);
}
#[test] #[test]
fn build_sync_decision_keeps_a_locally_modified_database() { fn build_sync_decision_keeps_a_locally_modified_database() {
// Same release metadata but a file matching no published build (local // Same release metadata but a file matching no published build (local
@ -1060,6 +1135,23 @@ mod tests {
assert_eq!(release_count, 1); assert_eq!(release_count, 1);
} }
#[test]
fn inspecting_an_empty_database_twice_sees_the_same_state() {
let dir = std::env::temp_dir().join(format!(
"idx-snapshot-empty-{}-{}",
std::process::id(),
fastrand::u64(..)
));
std::fs::create_dir_all(&dir).unwrap();
let db_path = dir.join("ownership.db");
std::fs::write(&db_path, b"").unwrap();
let first = super::inspect_local_db(&db_path).unwrap();
let second = super::inspect_local_db(&db_path).unwrap();
assert!(super::same_local_state(first.as_ref(), second.as_ref()));
let _ = std::fs::remove_dir_all(&dir);
}
#[test] #[test]
fn local_hash_covers_committed_changes_still_in_the_wal() { fn local_hash_covers_committed_changes_still_in_the_wal() {
let dir = std::env::temp_dir().join(format!( let dir = std::env::temp_dir().join(format!(

87
crates/idx-core/src/ownership/xlsx.rs
View file

@ -9,7 +9,7 @@
use std::collections::HashMap; use std::collections::HashMap;
use std::fs; use std::fs;
use std::io::{Cursor, Read}; use std::io::Cursor;
use std::path::Path; use std::path::Path;
use chrono::{Duration, NaiveDate}; use chrono::{Duration, NaiveDate};
@ -83,11 +83,7 @@ fn read_zip_entry(
))); )));
} }
}; };
let mut output = String::new(); super::read_zip_entry_bounded(&mut file, &format!("ownership XLSX entry {name}")).map(Some)
file.read_to_string(&mut output).map_err(|e| {
IdxError::ParseError(format!("failed to decode ownership XLSX entry {name}: {e}"))
})?;
Ok(Some(output))
} }
fn first_worksheet_name(zip: &mut zip::ZipArchive<Cursor<&[u8]>>) -> Result<String, IdxError> { fn first_worksheet_name(zip: &mut zip::ZipArchive<Cursor<&[u8]>>) -> Result<String, IdxError> {
@ -131,7 +127,10 @@ fn parse_shared_strings(xml: &str) -> Result<Vec<String>, IdxError> {
b"rPh" => in_phonetic = true, b"rPh" => in_phonetic = true,
_ => {} _ => {}
}, },
Ok(Event::Empty(e)) if e.local_name().as_ref() == b"si" => strings.push(String::new()), Ok(Event::Empty(e)) if e.local_name().as_ref() == b"si" => {
check_shared_string_count(strings.len())?;
strings.push(String::new())
}
Ok(Event::Text(text)) if in_text && !in_phonetic => { Ok(Event::Text(text)) if in_text && !in_phonetic => {
if let Some(value) = current.as_mut() { if let Some(value) = current.as_mut() {
let unescaped = text let unescaped = text
@ -141,7 +140,10 @@ fn parse_shared_strings(xml: &str) -> Result<Vec<String>, IdxError> {
} }
} }
Ok(Event::End(e)) => match e.local_name().as_ref() { Ok(Event::End(e)) => match e.local_name().as_ref() {
b"si" => strings.push(current.take().unwrap_or_default()), b"si" => {
check_shared_string_count(strings.len())?;
strings.push(current.take().unwrap_or_default())
}
b"t" => in_text = false, b"t" => in_text = false,
b"rPh" => in_phonetic = false, b"rPh" => in_phonetic = false,
_ => {} _ => {}
@ -167,15 +169,49 @@ struct SheetRow {
fn row_number(element: &BytesStart<'_>, fallback: usize) -> Result<usize, IdxError> { fn row_number(element: &BytesStart<'_>, fallback: usize) -> Result<usize, IdxError> {
for attribute in element.attributes() { for attribute in element.attributes() {
let attribute = attribute.map_err(|e| xml_error("worksheet", e))?; let attribute = attribute.map_err(|e| xml_error("worksheet", e))?;
if attribute.key.local_name().as_ref() == b"r" { if attribute.key.local_name().as_ref() == b"r"
if let Ok(number) = String::from_utf8_lossy(&attribute.value).trim().parse() { && let Ok(number) = String::from_utf8_lossy(&attribute.value).trim().parse()
return Ok(number); {
} return Ok(number);
} }
} }
Ok(fallback) Ok(fallback)
} }
/// Bounds on one worksheet. Real above-1% registers hold about 7,000 rows
/// and a few MB of text.
const MAX_CELL_TEXT_BYTES: usize = 256 * 1024 * 1024;
const MAX_SHEET_ROWS: usize = 1_000_000;
fn text_cap_error() -> IdxError {
xml_error(
"worksheet",
format!("cell text exceeds {MAX_CELL_TEXT_BYTES} bytes"),
)
}
const MAX_SHARED_STRINGS: usize = 2_000_000;
fn check_shared_string_count(count: usize) -> Result<(), IdxError> {
if count >= MAX_SHARED_STRINGS {
return Err(xml_error(
"shared strings",
format!("more than {MAX_SHARED_STRINGS} entries"),
));
}
Ok(())
}
fn check_row_count(rows: usize) -> Result<(), IdxError> {
if rows >= MAX_SHEET_ROWS {
return Err(xml_error(
"worksheet",
format!("more than {MAX_SHEET_ROWS} rows"),
));
}
Ok(())
}
fn parse_sheet_rows(xml: &str, shared_strings: &[String]) -> Result<Vec<SheetRow>, IdxError> { fn parse_sheet_rows(xml: &str, shared_strings: &[String]) -> Result<Vec<SheetRow>, IdxError> {
let mut reader = Reader::from_str(xml); let mut reader = Reader::from_str(xml);
let mut rows = Vec::new(); let mut rows = Vec::new();
@ -183,11 +219,13 @@ fn parse_sheet_rows(xml: &str, shared_strings: &[String]) -> Result<Vec<SheetRow
let mut cell: Option<(String, String)> = None; let mut cell: Option<(String, String)> = None;
let mut value = String::new(); let mut value = String::new();
let mut in_value = false; let mut in_value = false;
let mut cell_text_bytes = 0usize;
loop { loop {
match reader.read_event() { match reader.read_event() {
Ok(Event::Start(e)) => match e.local_name().as_ref() { Ok(Event::Start(e)) => match e.local_name().as_ref() {
b"row" => { b"row" => {
check_row_count(rows.len())?;
row = Some(SheetRow { row = Some(SheetRow {
number: row_number(&e, rows.len() + 1)?, number: row_number(&e, rows.len() + 1)?,
cells: HashMap::new(), cells: HashMap::new(),
@ -200,10 +238,13 @@ fn parse_sheet_rows(xml: &str, shared_strings: &[String]) -> Result<Vec<SheetRow
b"v" | b"t" if cell.is_some() => in_value = true, b"v" | b"t" if cell.is_some() => in_value = true,
_ => {} _ => {}
}, },
Ok(Event::Empty(e)) if e.local_name().as_ref() == b"row" => rows.push(SheetRow { Ok(Event::Empty(e)) if e.local_name().as_ref() == b"row" => {
number: row_number(&e, rows.len() + 1)?, check_row_count(rows.len())?;
cells: HashMap::new(), rows.push(SheetRow {
}), number: row_number(&e, rows.len() + 1)?,
cells: HashMap::new(),
})
}
Ok(Event::Text(text)) if in_value => { Ok(Event::Text(text)) if in_value => {
let unescaped = text.unescape().map_err(|e| xml_error("worksheet", e))?; let unescaped = text.unescape().map_err(|e| xml_error("worksheet", e))?;
value.push_str(&unescaped); value.push_str(&unescaped);
@ -219,12 +260,18 @@ fn parse_sheet_rows(xml: &str, shared_strings: &[String]) -> Result<Vec<SheetRow
let index = value.trim().parse::<usize>().map_err(|e| { let index = value.trim().parse::<usize>().map_err(|e| {
xml_error("worksheet", format!("bad shared string index: {e}")) xml_error("worksheet", format!("bad shared string index: {e}"))
})?; })?;
shared_strings.get(index).cloned().ok_or_else(|| { let shared = shared_strings.get(index).ok_or_else(|| {
xml_error( xml_error(
"worksheet", "worksheet",
format!("shared string {index} out of range"), format!("shared string {index} out of range"),
) )
})? })?;
// Check before copying: one large string may be
// referenced from many cells.
if cell_text_bytes + shared.len() > MAX_CELL_TEXT_BYTES {
return Err(text_cap_error());
}
shared.clone()
} }
// Excel stores the share code TRUE (TRINITI DINAMIK) // Excel stores the share code TRUE (TRINITI DINAMIK)
// as a boolean cell holding 1. // as a boolean cell holding 1.
@ -235,6 +282,10 @@ fn parse_sheet_rows(xml: &str, shared_strings: &[String]) -> Result<Vec<SheetRow
}, },
_ => value.clone(), _ => value.clone(),
}; };
cell_text_bytes += text.len();
if cell_text_bytes > MAX_CELL_TEXT_BYTES {
return Err(text_cap_error());
}
current_row.cells.insert(column, text); current_row.cells.insert(column, text);
} }
} }

5
crates/idx-mcp/Cargo.toml
View file

@ -3,8 +3,7 @@ name = "idx-mcp"
description = "MCP server for Indonesian stock market (IDX) market, financial, and ownership data" description = "MCP server for Indonesian stock market (IDX) market, financial, and ownership data"
version.workspace = true version.workspace = true
edition.workspace = true edition.workspace = true
# rmcp requires 1.88. rust-version.workspace = true
rust-version = "1.88"
license.workspace = true license.workspace = true
repository.workspace = true repository.workspace = true
homepage.workspace = true homepage.workspace = true
@ -24,7 +23,7 @@ rmcp = { version = "3.5", default-features = false, features = ["server", "macro
schemars = "1" schemars = "1"
serde.workspace = true serde.workspace = true
serde_json.workspace = true serde_json.workspace = true
tokio = { version = "1", features = ["macros", "rt-multi-thread", "io-std"] } tokio = { version = "1", features = ["macros", "rt-multi-thread", "io-std", "sync"] }
[dev-dependencies] [dev-dependencies]
# Tests drive the binary with the mock provider; release builds exclude it. # Tests drive the binary with the mock provider; release builds exclude it.

8
crates/idx-mcp/src/params.rs
View file

@ -8,7 +8,7 @@ use serde::Deserialize;
#[derive(Debug, Deserialize, JsonSchema)] #[derive(Debug, Deserialize, JsonSchema)]
pub struct SymbolsParams { pub struct SymbolsParams {
/// IDX tickers, e.g. ["BBCA", "BBRI"]; the `.JK` suffix is optional. /// IDX tickers, e.g. ["BBCA", "BBRI"]; the `.JK` suffix is optional. At most 20.
pub symbols: Vec<String>, pub symbols: Vec<String>,
} }
@ -36,7 +36,7 @@ pub struct ScreenParams {
#[serde(default)] #[serde(default)]
#[schemars(schema_with = "region_schema")] #[schemars(schema_with = "region_schema")]
pub region: Option<String>, pub region: Option<String>,
/// Maximum results (default 50). /// Maximum results (default 50, at most 200).
#[serde(default)] #[serde(default)]
pub limit: Option<usize>, pub limit: Option<usize>,
} }
@ -130,7 +130,7 @@ pub struct EntityParams {
pub struct SearchParams { pub struct SearchParams {
/// Full-text query over investor names. /// Full-text query over investor names.
pub query: String, pub query: String,
/// Maximum results (default 20). /// Maximum results (default 20, at most 100).
#[serde(default)] #[serde(default)]
pub limit: Option<usize>, pub limit: Option<usize>,
} }
@ -144,7 +144,7 @@ pub struct ChangesParams {
/// Only this ticker's changes. /// Only this ticker's changes.
#[serde(default)] #[serde(default)]
pub ticker: Option<String>, pub ticker: Option<String>,
/// Maximum rows returned (default 200); `total` reports the full count. /// Maximum rows returned (default 200, at most 2000); `total` reports the full count.
#[serde(default)] #[serde(default)]
pub limit: Option<usize>, pub limit: Option<usize>,
} }

109
crates/idx-mcp/src/server.rs
View file

@ -15,6 +15,7 @@ use rmcp::model::{CallToolResult, Implementation, ServerCapabilities, ServerConf
use rmcp::{ErrorData, ServerHandler, tool, tool_handler, tool_router}; use rmcp::{ErrorData, ServerHandler, tool, tool_handler, tool_router};
use serde::Serialize; use serde::Serialize;
use serde_json::{Map, Value, json}; use serde_json::{Map, Value, json};
use tokio::sync::Semaphore;
use crate::params::{ use crate::params::{
AnalysisKind, AnalyzeParams, ChangesParams, CompanyParams, CompanySection, EntityParams, AnalysisKind, AnalyzeParams, ChangesParams, CompanyParams, CompanySection, EntityParams,
@ -26,6 +27,15 @@ const NEWS_LIMIT: usize = 10;
const SCREEN_LIMIT: usize = 50; const SCREEN_LIMIT: usize = 50;
const SEARCH_LIMIT: usize = 20; const SEARCH_LIMIT: usize = 20;
const CHANGES_LIMIT: usize = 200; const CHANGES_LIMIT: usize = 200;
// Hard caps on what one call may request, so a misbehaving or prompt-injected
// agent cannot turn a call into unbounded upstream traffic or output.
const MAX_SYMBOLS: usize = 20;
const MAX_SCREEN_LIMIT: usize = 200;
const MAX_SEARCH_LIMIT: usize = 100;
const MAX_CHANGES_LIMIT: usize = 2000;
/// Tool calls running at once; the rest wait their turn.
static CALL_SLOTS: Semaphore = Semaphore::const_new(4);
const INSTRUCTIONS: &str = "Indonesian stock market (IDX) data. Market and financial \ const INSTRUCTIONS: &str = "Indonesian stock market (IDX) data. Market and financial \
tools fetch live data (cached on disk) and wrap it as {data, source, fetched_at, stale}; \ tools fetch live data (cached on disk) and wrap it as {data, source, fetched_at, stale}; \
@ -98,10 +108,10 @@ impl IdxServer {
for one or more IDX tickers." for one or more IDX tickers."
)] )]
async fn get_quotes(&self, params: Parameters<SymbolsParams>) -> ToolResult { async fn get_quotes(&self, params: Parameters<SymbolsParams>) -> ToolResult {
let SymbolsParams { symbols } = params.0; let symbols = match checked_symbols(params.0.symbols) {
if let Some(err) = empty_input("symbols", &symbols) { Ok(symbols) => symbols,
return err; Err(err) => return err,
} };
self.live(move |idx| { self.live(move |idx| {
let quotes = idx.market().quotes(&symbols, CachePolicy::Default)?; let quotes = idx.market().quotes(&symbols, CachePolicy::Default)?;
Ok(json!({ "quotes": quotes })) Ok(json!({ "quotes": quotes }))
@ -143,7 +153,7 @@ for one or more IDX tickers."
.unwrap_or("top-performers") .unwrap_or("top-performers")
.parse::<ScreenFilter>()?, .parse::<ScreenFilter>()?,
region: region.as_deref().unwrap_or("id").parse::<ScreenRegion>()?, region: region.as_deref().unwrap_or("id").parse::<ScreenRegion>()?,
limit: limit.unwrap_or(SCREEN_LIMIT), limit: bounded_limit("limit", limit, SCREEN_LIMIT, MAX_SCREEN_LIMIT)?,
}; };
idx.market().screen(&query, CachePolicy::Default) idx.market().screen(&query, CachePolicy::Default)
}) })
@ -157,7 +167,8 @@ earnings, analyst sentiment, insights, and news. Sections that fail are listed u
)] )]
async fn get_company(&self, params: Parameters<CompanyParams>) -> ToolResult { async fn get_company(&self, params: Parameters<CompanyParams>) -> ToolResult {
let CompanyParams { symbol, sections } = params.0; let CompanyParams { symbol, sections } = params.0;
let sections = sections.unwrap_or_else(|| CompanySection::ALL.to_vec()); let mut sections = sections.unwrap_or_else(|| CompanySection::ALL.to_vec());
dedup_in_order(&mut sections);
if let Some(err) = empty_input("sections", &sections) { if let Some(err) = empty_input("sections", &sections) {
return err; return err;
} }
@ -200,7 +211,8 @@ are listed under `errors`; the call fails only if every analysis fails."
)] )]
async fn analyze_stock(&self, params: Parameters<AnalyzeParams>) -> ToolResult { async fn analyze_stock(&self, params: Parameters<AnalyzeParams>) -> ToolResult {
let AnalyzeParams { symbol, kinds } = params.0; let AnalyzeParams { symbol, kinds } = params.0;
let kinds = kinds.unwrap_or_else(|| AnalysisKind::ALL.to_vec()); let mut kinds = kinds.unwrap_or_else(|| AnalysisKind::ALL.to_vec());
dedup_in_order(&mut kinds);
if let Some(err) = empty_input("kinds", &kinds) { if let Some(err) = empty_input("kinds", &kinds) {
return err; return err;
} }
@ -232,10 +244,10 @@ are listed under `errors`; the call fails only if every analysis fails."
tickers. Tickers that fail are listed under `failures`." tickers. Tickers that fail are listed under `failures`."
)] )]
async fn compare_stocks(&self, params: Parameters<SymbolsParams>) -> ToolResult { async fn compare_stocks(&self, params: Parameters<SymbolsParams>) -> ToolResult {
let SymbolsParams { symbols } = params.0; let symbols = match checked_symbols(params.0.symbols) {
if let Some(err) = empty_input("symbols", &symbols) { Ok(symbols) => symbols,
return err; Err(err) => return err,
} };
self.live(move |idx| { self.live(move |idx| {
let comparison = idx.analysis().compare(&symbols, CachePolicy::Default)?; let comparison = idx.analysis().compare(&symbols, CachePolicy::Default)?;
let failures: Vec<Value> = comparison let failures: Vec<Value> = comparison
@ -273,7 +285,8 @@ with concentration metrics (top1, top3, HHI, free float)."
async fn search_investors(&self, params: Parameters<SearchParams>) -> ToolResult { async fn search_investors(&self, params: Parameters<SearchParams>) -> ToolResult {
let SearchParams { query, limit } = params.0; let SearchParams { query, limit } = params.0;
self.local(move |investor| { self.local(move |investor| {
let entities = investor.search(&query, limit.unwrap_or(SEARCH_LIMIT))?; let limit = bounded_limit("limit", limit, SEARCH_LIMIT, MAX_SEARCH_LIMIT)?;
let entities = investor.search(&query, limit)?;
Ok(json!({ "entities": entities })) Ok(json!({ "entities": entities }))
}) })
.await .await
@ -291,9 +304,10 @@ release dates, optionally for one ticker."
limit, limit,
} = params.0; } = params.0;
self.local(move |investor| { self.local(move |investor| {
let limit = bounded_limit("limit", limit, CHANGES_LIMIT, MAX_CHANGES_LIMIT)?;
let mut changes = investor.changes(&from, &to, ticker.as_deref())?; let mut changes = investor.changes(&from, &to, ticker.as_deref())?;
let total = changes.len(); let total = changes.len();
changes.truncate(limit.unwrap_or(CHANGES_LIMIT)); changes.truncate(limit);
Ok(json!({ "from": from, "to": to, "total": total, "changes": changes })) Ok(json!({ "from": from, "to": to, "total": total, "changes": changes }))
}) })
.await .await
@ -315,6 +329,7 @@ the local database."
async fn ownership_sync(&self, params: Parameters<SyncParams>) -> ToolResult { async fn ownership_sync(&self, params: Parameters<SyncParams>) -> ToolResult {
let SyncParams { force } = params.0; let SyncParams { force } = params.0;
let investor = Investor::new(&self.config); let investor = Investor::new(&self.config);
// idx-core's database lock keeps the swap away from open readers.
blocking(move || investor.sync(None, force)).await blocking(move || investor.sync(None, force)).await
} }
} }
@ -380,6 +395,10 @@ where
T: Serialize + Send + 'static, T: Serialize + Send + 'static,
F: FnOnce() -> Result<T, IdxError> + Send + 'static, F: FnOnce() -> Result<T, IdxError> + Send + 'static,
{ {
let _slot = CALL_SLOTS
.acquire()
.await
.map_err(|err| ErrorData::internal_error(err.to_string(), None))?;
let result = tokio::task::spawn_blocking(move || f().and_then(to_json)) let result = tokio::task::spawn_blocking(move || f().and_then(to_json))
.await .await
.map_err(|err| ErrorData::internal_error(err.to_string(), None))?; .map_err(|err| ErrorData::internal_error(err.to_string(), None))?;
@ -389,15 +408,67 @@ where
}) })
} }
fn input_error(message: String) -> ToolResult {
let err = IdxError::InvalidInput(message);
Ok(CallToolResult::structured_error(Value::Object(error_body(
&err,
))))
}
/// An explicitly empty list is an input error, as in the CLI, rather than an /// An explicitly empty list is an input error, as in the CLI, rather than an
/// empty success an agent could mistake for data. /// empty success an agent could mistake for data.
fn empty_input<T>(field: &str, values: &[T]) -> Option<ToolResult> { fn empty_input<T>(field: &str, values: &[T]) -> Option<ToolResult> {
values.is_empty().then(|| { values
let err = IdxError::InvalidInput(format!("{field} must not be empty")); .is_empty()
Ok(CallToolResult::structured_error(Value::Object(error_body( .then(|| input_error(format!("{field} must not be empty")))
&err, }
))))
}) /// Non-empty, deduplicated (case-insensitively, first wins), and capped.
fn checked_symbols(symbols: Vec<String>) -> Result<Vec<String>, ToolResult> {
if let Some(err) = empty_input("symbols", &symbols) {
return Err(err);
}
let mut seen = std::collections::HashSet::new();
let symbols: Vec<String> = symbols
.into_iter()
.filter(|symbol| seen.insert(symbol.trim().to_uppercase()))
.collect();
if symbols.len() > MAX_SYMBOLS {
return Err(input_error(format!(
"at most {MAX_SYMBOLS} symbols per call (got {})",
symbols.len()
)));
}
Ok(symbols)
}
/// Each section or analysis runs once, however often it is requested.
fn dedup_in_order<T: PartialEq + Copy>(values: &mut Vec<T>) {
let mut seen = Vec::with_capacity(values.len());
values.retain(|value| {
let new = !seen.contains(value);
if new {
seen.push(*value);
}
new
});
}
fn bounded_limit(
field: &str,
value: Option<usize>,
default: usize,
max: usize,
) -> Result<usize, IdxError> {
match value.unwrap_or(default) {
0 => Err(IdxError::InvalidInput(format!(
"{field} must be at least 1"
))),
n if n > max => Err(IdxError::InvalidInput(format!(
"{field} must be at most {max} (got {n})"
))),
n => Ok(n),
}
} }
fn to_json<T: Serialize>(value: T) -> Result<Value, IdxError> { fn to_json<T: Serialize>(value: T) -> Result<Value, IdxError> {

333
crates/idx-mcp/tests/golden/tools.json Normal file
View file

@ -0,0 +1,333 @@
[
{
"description": "Analyze one IDX ticker: technical signals (trend, RSI, MACD, moving averages) and fundamental growth, valuation, and risk ratings. Analyses that fail are listed under `errors`; the call fails only if every analysis fails.",
"inputSchema": {
"$defs": {
"AnalysisKind": {
"enum": [
"technical",
"growth",
"valuation",
"risk"
],
"type": "string"
}
},
"$schema": "https://json-schema.org/draft/2020-12/schema",
"properties": {
"kinds": {
"description": "Analyses to run (default: all).",
"items": {
"$ref": "#/$defs/AnalysisKind"
},
"type": [
"array",
"null"
]
},
"symbol": {
"description": "IDX ticker, e.g. \"BBCA\".",
"type": "string"
}
},
"required": [
"symbol"
],
"type": "object"
},
"name": "analyze_stock"
},
{
"description": "Compare fundamental reports (growth, valuation, risk) across IDX tickers. Tickers that fail are listed under `failures`.",
"inputSchema": {
"$schema": "https://json-schema.org/draft/2020-12/schema",
"properties": {
"symbols": {
"description": "IDX tickers, e.g. [\"BBCA\", \"BBRI\"]; the `.JK` suffix is optional. At most 20.",
"items": {
"type": "string"
},
"type": "array"
}
},
"required": [
"symbols"
],
"type": "object"
},
"name": "compare_stocks"
},
{
"description": "Company data for one IDX ticker: profile, financial statements, earnings, analyst sentiment, insights, and news. Sections that fail are listed under `errors`; the call fails only if every section fails.",
"inputSchema": {
"$defs": {
"CompanySection": {
"enum": [
"profile",
"financials",
"earnings",
"sentiment",
"insights",
"news"
],
"type": "string"
}
},
"$schema": "https://json-schema.org/draft/2020-12/schema",
"properties": {
"sections": {
"description": "Sections to fetch (default: all).",
"items": {
"$ref": "#/$defs/CompanySection"
},
"type": [
"array",
"null"
]
},
"symbol": {
"description": "IDX ticker, e.g. \"BBCA\".",
"type": "string"
}
},
"required": [
"symbol"
],
"type": "object"
},
"name": "get_company"
},
{
"description": "Every IDX holding (above 1%) of one investor, matched by name.",
"inputSchema": {
"$schema": "https://json-schema.org/draft/2020-12/schema",
"properties": {
"entity": {
"description": "Investor name or part of it, e.g. \"DWIMURIA\"; the closest match is used.",
"type": "string"
}
},
"required": [
"entity"
],
"type": "object"
},
"name": "get_investor_holdings"
},
{
"description": "Holder changes (new, exit, increase, decrease) between two KSEI release dates, optionally for one ticker.",
"inputSchema": {
"$schema": "https://json-schema.org/draft/2020-12/schema",
"properties": {
"from": {
"description": "Earlier release date, YYYY-MM-DD (see ownership_status).",
"type": "string"
},
"limit": {
"default": null,
"description": "Maximum rows returned (default 200, at most 2000); `total` reports the full count.",
"format": "uint",
"minimum": 0,
"type": [
"integer",
"null"
]
},
"ticker": {
"default": null,
"description": "Only this ticker's changes.",
"type": [
"string",
"null"
]
},
"to": {
"description": "Later release date, YYYY-MM-DD.",
"type": "string"
}
},
"required": [
"from",
"to"
],
"type": "object"
},
"name": "get_ownership_changes"
},
{
"description": "OHLCV price history for one IDX ticker.",
"inputSchema": {
"$schema": "https://json-schema.org/draft/2020-12/schema",
"properties": {
"interval": {
"default": null,
"description": "Candle size (default \"1d\").",
"enum": [
"1d",
"1wk",
"1mo"
],
"type": "string"
},
"period": {
"default": null,
"description": "Lookback window (default \"3mo\").",
"enum": [
"1d",
"5d",
"1mo",
"3mo",
"6mo",
"1y",
"2y",
"5y"
],
"type": "string"
},
"symbol": {
"description": "IDX ticker, e.g. \"BBCA\".",
"type": "string"
}
},
"required": [
"symbol"
],
"type": "object"
},
"name": "get_price_history"
},
{
"description": "Latest quotes (price, change, volume, 52-week range, market cap) for one or more IDX tickers.",
"inputSchema": {
"$schema": "https://json-schema.org/draft/2020-12/schema",
"properties": {
"symbols": {
"description": "IDX tickers, e.g. [\"BBCA\", \"BBRI\"]; the `.JK` suffix is optional. At most 20.",
"items": {
"type": "string"
},
"type": "array"
}
},
"required": [
"symbols"
],
"type": "object"
},
"name": "get_quotes"
},
{
"description": "Shareholders above 1% of an IDX ticker from the latest KSEI release, with concentration metrics (top1, top3, HHI, free float).",
"inputSchema": {
"$schema": "https://json-schema.org/draft/2020-12/schema",
"properties": {
"ticker": {
"description": "IDX ticker, e.g. \"BBCA\".",
"type": "string"
}
},
"required": [
"ticker"
],
"type": "object"
},
"name": "get_shareholders"
},
{
"description": "Ownership database status: whether it has data, and the imported KSEI release dates.",
"inputSchema": {
"properties": {},
"type": "object"
},
"name": "ownership_status"
},
{
"description": "Download or refresh the KSEI ownership snapshot (several MB) into the local database.",
"inputSchema": {
"$schema": "https://json-schema.org/draft/2020-12/schema",
"properties": {
"force": {
"default": false,
"description": "Re-download even when the local snapshot is current.",
"type": "boolean"
}
},
"type": "object"
},
"name": "ownership_sync"
},
{
"description": "Run a stock screen (top/worst performers, high dividend, low P/E, 52-week highs/lows, high volume, large cap).",
"inputSchema": {
"$schema": "https://json-schema.org/draft/2020-12/schema",
"properties": {
"filter": {
"default": null,
"description": "Screen to run (default \"top-performers\").",
"enum": [
"top-performers",
"worst-performers",
"high-dividend",
"low-pe",
"52w-high",
"52w-low",
"high-volume",
"large-cap"
],
"type": "string"
},
"limit": {
"default": null,
"description": "Maximum results (default 50, at most 200).",
"format": "uint",
"minimum": 0,
"type": [
"integer",
"null"
]
},
"region": {
"default": null,
"description": "Market region (default \"id\").",
"enum": [
"id",
"us",
"sg",
"hk",
"jp"
],
"type": "string"
}
},
"type": "object"
},
"name": "screen_stocks"
},
{
"description": "Search investor names in the KSEI ownership database.",
"inputSchema": {
"$schema": "https://json-schema.org/draft/2020-12/schema",
"properties": {
"limit": {
"default": null,
"description": "Maximum results (default 20, at most 100).",
"format": "uint",
"minimum": 0,
"type": [
"integer",
"null"
]
},
"query": {
"description": "Full-text query over investor names.",
"type": "string"
}
},
"required": [
"query"
],
"type": "object"
},
"name": "search_investors"
}
]

41
crates/idx-mcp/tests/mcp.rs
View file

@ -181,6 +181,31 @@ fn lists_grouped_tools_and_hides_sync_unless_allowed() {
assert_eq!(client.tool_names(), expected); assert_eq!(client.tool_names(), expected);
} }
/// What agents see (names, descriptions, input schemas) is part of the
/// contract: changes must show up as a reviewed diff. Regenerate with
/// `UPDATE_GOLDEN=1`.
#[test]
fn tool_definitions_match_golden() {
let mut client = Client::start(&["--allow-sync"], false);
let mut tools = client.request("tools/list", json!({}))["tools"]
.as_array()
.unwrap()
.clone();
tools.sort_by(|a, b| a["name"].as_str().cmp(&b["name"].as_str()));
let rendered = serde_json::to_string_pretty(&tools).unwrap() + "\n";
let golden = Path::new(env!("CARGO_MANIFEST_DIR")).join("tests/golden/tools.json");
if std::env::var_os("UPDATE_GOLDEN").is_some() {
std::fs::create_dir_all(golden.parent().unwrap()).unwrap();
std::fs::write(&golden, &rendered).unwrap();
}
let expected = std::fs::read_to_string(&golden).expect("tests/golden/tools.json");
assert!(
rendered == expected,
"tool definitions changed; review and rerun with UPDATE_GOLDEN=1:\n{rendered}"
);
}
#[test] #[test]
fn live_tools_wrap_data_with_freshness() { fn live_tools_wrap_data_with_freshness() {
let mut client = Client::start(&[], false); let mut client = Client::start(&[], false);
@ -254,6 +279,22 @@ fn errors_carry_cli_codes() {
assert_eq!(body["code"], "INVALIDINPUT", "{tool}"); assert_eq!(body["code"], "INVALIDINPUT", "{tool}");
} }
let many: Vec<String> = (0..21).map(|i| format!("S{i}")).collect();
for (tool, arguments) in [
("get_quotes", json!({"symbols": many})),
("screen_stocks", json!({"limit": 201})),
("screen_stocks", json!({"limit": 0})),
("get_quotes", json!({"symbols": ["../../REPORT"]})),
] {
let (is_error, body) = client.call(tool, arguments.clone());
assert!(is_error, "{tool} {arguments}: {body}");
assert_eq!(body["code"], "INVALIDINPUT", "{tool} {arguments}");
}
// Duplicates collapse instead of counting against the cap.
let repeated = vec!["BBCA"; 50];
let quotes = client.ok("get_quotes", json!({"symbols": repeated}));
assert_eq!(quotes["quotes"].as_array().unwrap().len(), 1);
// Empty ownership database: investor tools point at the sync path. // Empty ownership database: investor tools point at the sync path.
let status = client.ok("ownership_status", json!({})); let status = client.ok("ownership_status", json!({}));
assert_eq!(status["populated"], false); assert_eq!(status["populated"], false);

40
docs/SMOKE.md
View file

@ -8,7 +8,7 @@ offline with pinned tools:
| Check | Covers | | Check | Covers |
| --- | --- | | --- | --- |
| `fmt`, `clippy` | formatting and lints (`--all-targets`) | | `fmt`, `clippy` | formatting and lints (`--all-targets`) |
| `test` | unit/module tests, `tests/cli.rs` CLI integration tests, and `tests/golden.rs` output parity | | `test` | unit/module tests, `tests/cli.rs` CLI integration tests, `tests/golden.rs` output parity, and `crates/idx-mcp/tests/mcp.rs` MCP stdio tests |
| `default`, `package` | release build and `cargo package` verification | | `default`, `package` | release build and `cargo package` verification |
| `packaged-cli` | the wrapped Nix release binary emits JSON, generates completions, and returns a JSON `CACHEMISS` error with exit 3 offline | | `packaged-cli` | the wrapped Nix release binary emits JSON, generates completions, and returns a JSON `CACHEMISS` error with exit 3 offline |
| `ownership-publisher`, `publisher-test` | packaged publisher entry point and its regression scenarios | | `ownership-publisher`, `publisher-test` | packaged publisher entry point and its regression scenarios |
@ -25,6 +25,11 @@ user-visible output changed. Regenerate with
`UPDATE_GOLDEN=1 cargo test --test golden` only when the change is intended, `UPDATE_GOLDEN=1 cargo test --test golden` only when the change is intended,
and commit the golden diff with the reason. and commit the golden diff with the reason.
`crates/idx-mcp/tests/mcp.rs` drives `idx-mcp` over stdio JSON-RPC with the
mock provider. `crates/idx-mcp/tests/golden/tools.json` pins what agents see (tool
names, descriptions, input schemas); regenerate it the same way after an
intended change.
For a focused loop, run one check or filter Cargo tests: For a focused loop, run one check or filter Cargo tests:
```bash ```bash
@ -66,3 +71,36 @@ and keeps per-case `.stdout`, `.stderr`, and `.validation` logs there.
`scripts/audit-msn-fundamentals.sh` is a heavier sweep of valuation output `scripts/audit-msn-fundamentals.sh` is a heavier sweep of valuation output
across the whole IDX MSN symbol map, for provider-health audits. across the whole IDX MSN symbol map, for provider-health audits.
## Agent smoke test
The MCP tests prove the protocol works; they cannot show whether a real
coding agent picks the right tools. `tests/agents/agent-smoke.sh` runs each
case in `tests/agents/cases.json` through Claude Code, Codex, pi, Hermes Agent,
and opencode in headless mode against a local `idx-mcp` build, then checks
the tools each agent called (and their key arguments) and, with mock data, key
facts in the answer. It is opt-in: it needs agent logins and model credits,
and agent behavior varies between runs, so it is never part of the gate.
```bash
nix develop .#agents --command tests/agents/agent-smoke.sh # every agent found
nix develop .#agents --command tests/agents/agent-smoke.sh --agents claude,codex --runs 3
tests/agents/agent-smoke.sh --live --case shareholders --keep # real data
```
- Each agent gets an isolated config that registers only the local server;
your own agent settings and MCP servers are not loaded or changed. Logins
are copied from the agent's home (Codex `auth.json`, pi and Hermes auth
files) into a temp dir that is deleted afterwards; `--keep` keeps the run
output but still deletes every copied agent config.
- Tools are auto-approved for the run (Claude `dontAsk` with
`mcp__idx__*`, Codex `default_tools_approval_mode="approve"`); Claude and pi
also get no built-in tools, so they can only answer through MCP.
- `IDX_AGENT_MODEL_<AGENT>` picks a model (for example
`IDX_AGENT_MODEL_CLAUDE=sonnet`). `IDX_AGENT_EXEC_PREFIX` prefixes every
agent command, for hosts that inject API keys through a wrapper.
- The run fails only when an agent fails a case in every run; use `--runs` to
measure pass rates. `--keep` keeps raw agent output and `results.jsonl`.
- `nix develop .#agents` adds Claude Code, Codex, and opencode from nixpkgs
after your PATH, so installed versions win. pi and Hermes Agent must be
installed on the host.

18
flake.nix
View file

@ -280,6 +280,7 @@
shellcheck -s sh ${installerSrc}/install.sh shellcheck -s sh ${installerSrc}/install.sh
shellcheck ${installerSrc}/scripts/install-sh-test.sh shellcheck ${installerSrc}/scripts/install-sh-test.sh
shellcheck ${./scripts/live-smoke.sh} shellcheck ${./scripts/live-smoke.sh}
shellcheck ${./tests/agents/agent-smoke.sh}
touch "$out" touch "$out"
''; '';
publisher-test = pkgs.runCommand "idx-publisher-test" { publisher-test = pkgs.runCommand "idx-publisher-test" {
@ -296,6 +297,23 @@
''; '';
}; };
# `nix develop .#agents`: the default shell plus coding-agent CLIs for
# tests/agents/agent-smoke.sh. They are appended to PATH, so agents
# already installed (and logged in) on the host take precedence.
# pi and Hermes Agent are not in the pinned nixpkgs; use host installs.
devShells.agents =
let
agentPkgs = import nixpkgs {
inherit system;
config.allowUnfreePredicate = pkg: lib.getName pkg == "claude-code";
};
in
self.devShells.${system}.default.overrideAttrs (old: {
shellHook = (old.shellHook or "") + ''
export PATH="$PATH:${lib.makeBinPath (with agentPkgs; [ claude-code codex opencode ])}"
'';
});
devShells.default = pkgs.mkShell { devShells.default = pkgs.mkShell {
inputsFrom = [ idxPackage ]; inputsFrom = [ idxPackage ];
packages = with pkgs; [ packages = with pkgs; [

4
scripts/publish-ownership-snapshot.sh
View file

@ -464,7 +464,9 @@ MANIFEST_PATH="$OUTPUT_DIR/ownership-snapshot-manifest.json"
# A rebuild for the published as-of date (e.g. after a parser fix) lists the # A rebuild for the published as-of date (e.g. after a parser fix) lists the
# builds it replaces, so clients still holding one update without --force. # builds it replaces, so clients still holding one update without --force.
if [[ "$BUILT_AS_OF" == "$PUBLISHED_AS_OF" && "$PUBLISHED_SHA256" =~ ^[0-9a-fA-F]{64}$ ]]; then # Only when it keeps at least the published history: clients never downgrade.
if [[ "$BUILT_AS_OF" == "$PUBLISHED_AS_OF" && "$PUBLISHED_SHA256" =~ ^[0-9a-fA-F]{64}$ ]] \
&& [[ "$PUBLISHED_RELEASE_COUNT" =~ ^[0-9]+$ ]] && (( BUILT_RELEASE_COUNT >= PUBLISHED_RELEASE_COUNT )); then
replaces_tmp="$PUBLISH_WORKDIR/manifest-replaces.json" replaces_tmp="$PUBLISH_WORKDIR/manifest-replaces.json"
jq --arg published "$PUBLISHED_SHA256" \ jq --arg published "$PUBLISHED_SHA256" \
--argjson previous "$(jq -c '.snapshot.replaces_sqlite_sha256 // []' <<< "$PUBLISHED_MANIFEST" 2>/dev/null || echo '[]')" ' --argjson previous "$(jq -c '.snapshot.replaces_sqlite_sha256 // []' <<< "$PUBLISHED_MANIFEST" 2>/dev/null || echo '[]')" '

316
tests/agents/agent-smoke.sh Executable file
View file

@ -0,0 +1,316 @@
#!/usr/bin/env bash
# Agent smoke test: runs real coding agents headless against a local idx-mcp
# build and checks which MCP tools they call. Opt-in: it needs agent logins
# and model credits, and agent behavior is nondeterministic. It is not part
# of `nix flake check`. See docs/SMOKE.md.
set -Eeuo pipefail
usage() {
cat <<'EOF'
Usage: tests/agents/agent-smoke.sh [options]
Runs each case in tests/agents/cases.json through each agent, with an
isolated agent config that registers only the local idx-mcp, and checks the
tools called (and, with mock data, key facts in the answer).
Options:
--agents <list> Comma-separated: claude,codex,pi,hermes,opencode
(default: every one found on PATH)
--case <id> Run only this case (repeatable)
--runs <n> Runs per agent and case (default: 1)
--live Real upstream data and the published ownership snapshot
instead of mock fixtures; answer text is not checked
--keep Keep the work directory (raw agent output, results.jsonl)
--help Show this help
Environment:
IDX_AGENT_EXEC_PREFIX Command prefix for every agent, e.g. a secrets
wrapper: "with-ai-secrets ai-tools"
IDX_AGENT_MODEL_<AGENT> Model override, e.g. IDX_AGENT_MODEL_CLAUDE=sonnet,
IDX_AGENT_MODEL_OPENCODE=opencode-go/glm-5.3
IDX_AGENT_TIMEOUT Seconds per agent run (default: 300)
Exit status: 0 when every agent passed every case at least once, 1 when some
agent failed a case in all runs, 2 on usage or setup errors.
EOF
}
repo_root="$(cd "$(dirname "${BASH_SOURCE[0]}")/../.." && pwd)"
cases_file="$repo_root/tests/agents/cases.json"
all_agents=(claude codex pi hermes opencode)
agents=()
only_cases=()
runs=1
live=0
keep=0
timeout_secs="${IDX_AGENT_TIMEOUT:-300}"
read -r -a exec_prefix <<< "${IDX_AGENT_EXEC_PREFIX:-}"
while [[ $# -gt 0 ]]; do
case "$1" in
--agents) IFS=, read -r -a agents <<< "${2:?}"; shift 2 ;;
--case) only_cases+=("${2:?}"); shift 2 ;;
--runs) runs="${2:?}"; shift 2 ;;
--live) live=1; shift ;;
--keep) keep=1; shift ;;
--help|-h) usage; exit 0 ;;
*) usage >&2; exit 2 ;;
esac
done
[[ "$runs" =~ ^[1-9][0-9]*$ ]] || { echo "--runs must be a positive integer" >&2; exit 2; }
if ((BASH_VERSINFO[0] < 4 || (BASH_VERSINFO[0] == 4 && BASH_VERSINFO[1] < 4))); then
echo "agent-smoke: bash 4.4+ is required (run it inside nix develop)" >&2
exit 2
fi
for tool in jq cargo timeout; do
command -v "$tool" >/dev/null || { echo "agent-smoke: $tool is required" >&2; exit 2; }
done
if [[ ${#agents[@]} -eq 0 ]]; then
for agent in "${all_agents[@]}"; do
command -v "$agent" >/dev/null && agents+=("$agent")
done
fi
[[ ${#agents[@]} -gt 0 ]] || { echo "agent-smoke: no agent CLI found on PATH" >&2; exit 2; }
for agent in "${agents[@]}"; do
[[ " ${all_agents[*]} " == *" $agent "* ]] || { echo "unknown agent: $agent" >&2; exit 2; }
command -v "$agent" >/dev/null || { echo "agent-smoke: $agent not found on PATH" >&2; exit 2; }
done
work="$(mktemp -d "${TMPDIR:-/tmp}/idx-agent-smoke.XXXXXX")"
cleanup() {
if [[ "$keep" == 1 ]]; then
# Never leave copied agent logins or settings (which may hold API keys)
# behind; each run's raw output is already under runs/.
rm -rf -- "$work/codex" "$work/pi" "$work/hermes" "$work/opencode"
echo "agent-smoke: kept $work (copied agent configs removed)"
elif [[ "$work" == */idx-agent-smoke.* ]]; then
rm -rf -- "$work"
fi
}
trap cleanup EXIT
# --- idx-mcp under test ----------------------------------------------------
echo "agent-smoke: building idx-mcp ($([[ $live == 1 ]] && echo live || echo mock) data)"
features=()
[[ "$live" == 1 ]] || features=(--features idx-core/testing)
(cd "$repo_root" && cargo build -q --locked -p idx-cli -p idx-mcp "${features[@]}")
idx="$repo_root/target/debug/idx"
data_env=(
"XDG_CONFIG_HOME=$work/idx/config"
"XDG_CACHE_HOME=$work/idx/cache"
"XDG_DATA_HOME=$work/idx/data"
)
if [[ "$live" == 1 ]]; then
env "${data_env[@]}" "$idx" -o json ownership sync >/dev/null
else
data_env+=(IDX_USE_MOCK_PROVIDER=1)
fixtures="$repo_root/crates/idx-core/tests/fixtures"
# The importer derives the release from the IDX file name.
cp "$fixtures/ksei_above1_20260831_excerpt.xlsx" "$work/peng-2026-08-00017-satu-persen.xlsx"
for file in "$work/peng-2026-08-00017-satu-persen.xlsx" \
"$fixtures/ksei_balancepos_20260130_excerpt.txt" \
"$fixtures/ksei_balancepos_20260227_excerpt.txt"; do
env "${data_env[@]}" "$idx" -o json ownership import --file "$file" >/dev/null
done
fi
# One launcher for every agent: isolated data dirs, and the idx-core crate
# as working directory (the mock provider reads tests/fixtures from there).
launcher="$work/idx-mcp"
{
echo '#!/usr/bin/env bash'
printf 'export %q\n' "${data_env[@]}"
printf 'cd %q\n' "$repo_root/crates/idx-core"
printf 'exec %q "$@"\n' "$repo_root/target/debug/idx-mcp"
} > "$launcher"
chmod +x "$launcher"
# --- per-agent setup, run, and tool-call extraction ---------------------------
# Each run_<agent> writes raw output to $out.raw, the answer to $out.answer, and
# one {"name","args"} JSON object per idx tool call to $out.calls, with names
# normalized to the bare tool name.
copy_if_present() { [[ -f "$1" ]] && cp "$1" "$2/"; return 0; }
setup_claude() {
printf '{"mcpServers":{"idx":{"type":"stdio","command":"%s","args":[]}}}\n' "$launcher" > "$work/claude-mcp.json"
}
run_claude() {
local prompt="$1" out="$2" model=()
[[ -n "${IDX_AGENT_MODEL_CLAUDE:-}" ]] && model=(--model "$IDX_AGENT_MODEL_CLAUDE")
"${exec_prefix[@]}" claude -p "$prompt" "${model[@]}" \
--mcp-config "$work/claude-mcp.json" --strict-mcp-config \
--setting-sources project --no-session-persistence --tools "" \
--permission-mode dontAsk --allowedTools "mcp__idx__*" \
--output-format stream-json --verbose --max-turns 10 < /dev/null > "$out.raw"
jq -c 'select(.type=="assistant") | .message.content[]? | select(.type=="tool_use")
| select(.name | startswith("mcp__idx__"))
| {name: (.name | ltrimstr("mcp__idx__")), args: .input}' "$out.raw" > "$out.calls"
jq -r 'select(.type=="result") | .result // ""' "$out.raw" > "$out.answer"
}
setup_codex() {
mkdir -p "$work/codex"
copy_if_present "${CODEX_HOME:-$HOME/.codex}/auth.json" "$work/codex"
}
run_codex() {
local prompt="$1" out="$2" model=()
[[ -n "${IDX_AGENT_MODEL_CODEX:-}" ]] && model=(-m "$IDX_AGENT_MODEL_CODEX")
CODEX_HOME="$work/codex" "${exec_prefix[@]}" codex exec --json --ephemeral \
--skip-git-repo-check -s read-only "${model[@]}" \
-c "mcp_servers.idx.command=\"$launcher\"" -c 'mcp_servers.idx.args=[]' \
-c 'mcp_servers.idx.default_tools_approval_mode="approve"' \
"$prompt" < /dev/null > "$out.raw"
jq -c 'select(.type=="item.completed" and .item.type=="mcp_tool_call" and .item.server=="idx")
| select(.item.status=="completed") | {name: .item.tool, args: .item.arguments}' "$out.raw" > "$out.calls"
jq -r 'select(.type=="item.completed" and .item.type=="agent_message") | .item.text' "$out.raw" > "$out.answer"
}
setup_pi() {
local src="${PI_CODING_AGENT_DIR:-$HOME/.pi/agent}"
mkdir -p "$work/pi"
for file in auth.json models.json settings.json; do copy_if_present "$src/$file" "$work/pi"; done
# "direct" exposes each MCP tool by name (the default "codemode" wraps them).
jq -n --arg cmd "$launcher" '{mcpServers: {idx: {command: $cmd, exposure: "direct"}}}' > "$work/pi/mcp.json"
}
run_pi() {
local prompt="$1" out="$2" model=()
[[ -n "${IDX_AGENT_MODEL_PI:-}" ]] && model=(--model "$IDX_AGENT_MODEL_PI")
# No built-in tools, so the agent can only reach the data through MCP.
PI_CODING_AGENT_DIR="$work/pi" "${exec_prefix[@]}" pi -p --mode json --no-session \
--no-builtin-tools --no-skills --no-context-files "${model[@]}" "$prompt" < /dev/null > "$out.raw"
jq -c 'select(.type=="tool_execution_start") | select(.toolName | startswith("mcp__idx__"))
| {name: (.toolName | ltrimstr("mcp__idx__")), args: .args}' "$out.raw" > "$out.calls"
jq -r 'select(.type=="agent_end") | .messages[-1].content[]? | select(.type=="text") | .text' "$out.raw" > "$out.answer"
}
setup_hermes() {
local src="${HERMES_HOME:-$HOME/.hermes}"
mkdir -p "$work/hermes"
copy_if_present "$src/auth.json" "$work/hermes"
copy_if_present "$src/.env" "$work/hermes"
{
# The top-level `model:` block, up to the next top-level key.
[[ -f "$src/config.yaml" ]] && awk '/^[^[:space:]#]/ { in_model = /^model:/ } in_model' "$src/config.yaml"
printf 'mcp_servers:\n idx:\n command: %s\n enabled: true\n' "$launcher"
} > "$work/hermes/config.yaml"
}
run_hermes() {
local prompt="$1" out="$2" model=() session
[[ -n "${IDX_AGENT_MODEL_HERMES:-}" ]] && model=(-m "$IDX_AGENT_MODEL_HERMES")
HERMES_HOME="$work/hermes" "${exec_prefix[@]}" hermes -z "$prompt" "${model[@]}" \
--usage-file "$out.usage.json" < /dev/null > "$out.answer"
# -z prints only the answer; tool calls come from the session export.
session="$(jq -r '.session_id // empty' "$out.usage.json")"
[[ -n "$session" ]] || return 1
HERMES_HOME="$work/hermes" hermes sessions export --format jsonl --session-id "$session" \
"$out.raw" --yes < /dev/null >/dev/null
# Hermes may call MCP tools directly or through its tool_call wrapper.
jq -c '.messages[]? | .tool_calls[]? | .function
| (.arguments | (try fromjson catch {})) as $args
| if .name == "tool_call" then ($args.calls[]? | {name, args: .arguments}) else {name, args: $args} end
| select(.name | startswith("mcp__idx__"))
| .name |= ltrimstr("mcp__idx__")' "$out.raw" > "$out.calls"
}
setup_opencode() {
mkdir -p "$work/opencode/cwd"
jq -n --arg cmd "$launcher" \
'{"$schema": "https://opencode.ai/config.json", mcp: {idx: {type: "local", command: [$cmd], enabled: true}}}' \
> "$work/opencode/cwd/opencode.json"
}
run_opencode() {
local prompt="$1" out="$2" model=() home="$work/opencode"
[[ -n "${IDX_AGENT_MODEL_OPENCODE:-}" ]] && model=(-m "$IDX_AGENT_MODEL_OPENCODE")
(cd "$home/cwd" && XDG_CONFIG_HOME="$home/config" XDG_DATA_HOME="$home/data" \
XDG_CACHE_HOME="$home/cache" XDG_STATE_HOME="$home/state" \
"${exec_prefix[@]}" opencode run --pure --format json "${model[@]}" "$prompt" < /dev/null > "$out.raw")
jq -c 'select(.type=="tool_use") | .part | select(.tool | startswith("idx_"))
| select(.state.status=="completed") | {name: (.tool | ltrimstr("idx_")), args: .state.input}' "$out.raw" > "$out.calls"
jq -r 'select(.type=="text") | .part.text' "$out.raw" > "$out.answer"
}
# --- checks -----------------------------------------------------------------
# Prints failure reasons for one run; no output means it passed.
check_run() {
local case_json="$1" out="$2"
jq -rn --argjson case "$case_json" --slurpfile calls "$out.calls" '
$case.expect_tools[] as $want
| if $want.any then
select([$calls[] | .name] | any(. as $n | $want.any | index($n)) | not)
| "no call to any of \($want.any | join(", "))"
else
([$calls[] | select(.name == $want.name)]) as $matches
| if ($matches | length) == 0 then "no call to \($want.name)"
elif ($want.args // {}) as $args | ($matches | any(.args | contains($args))) | not
then "\($want.name) called without \($want.args | tojson): \($matches | map(.args) | tojson)"
else empty end
end' || echo "could not parse the agent's tool calls (see $out.raw)"
if [[ "$live" == 0 ]]; then
local pattern
while IFS= read -r pattern; do
# jq regexes (Oniguruma) support (?i) on every platform, unlike grep -P.
jq -Rse --arg re "$pattern" 'test($re)' "$out.answer" >/dev/null \
|| echo "answer does not match /$pattern/"
done < <(jq -r '.expect_text[]?' <<< "$case_json")
fi
}
# --- run --------------------------------------------------------------------
mapfile -t case_ids < <(jq -r '.[].id' "$cases_file")
if [[ ${#only_cases[@]} -gt 0 ]]; then
for id in "${only_cases[@]}"; do
[[ " ${case_ids[*]} " == *" $id "* ]] || { echo "unknown case: $id" >&2; exit 2; }
done
case_ids=("${only_cases[@]}")
fi
results="$work/results.jsonl"
: > "$results"
echo "agent-smoke: agents=${agents[*]} cases=${#case_ids[@]} runs=$runs"
for agent in "${agents[@]}"; do
"setup_$agent"
for id in "${case_ids[@]}"; do
case_json="$(jq -c --arg id "$id" '.[] | select(.id == $id)' "$cases_file")"
prompt="$(jq -r '.prompt' <<< "$case_json") Use the idx MCP tools."
for ((run = 1; run <= runs; run++)); do
out="$work/runs/$agent/$id.$run"
mkdir -p "$(dirname "$out")"
: > "$out.calls"; : > "$out.answer"
started=$SECONDS
if timeout "$timeout_secs" bash -c "set -eo pipefail; $(declare -f "run_$agent"); $(declare -p work launcher exec_prefix); run_$agent \"\$1\" \"\$2\"" _ "$prompt" "$out" 2> "$out.stderr"; then
# Malformed agent output fails this run, not the whole matrix.
failures="$(check_run "$case_json" "$out")" \
|| failures="${failures:+$failures$'\n'}could not check the agent output (see $out.raw)"
else
failures="agent exited with status $? (see $out.stderr)"
fi
status=pass
[[ -z "$failures" ]] || status=fail
jq -nc --arg agent "$agent" --arg case "$id" --argjson run "$run" --arg status "$status" \
--arg failures "$failures" --argjson secs "$((SECONDS - started))" \
--slurpfile calls "$out.calls" \
'{agent: $agent, case: $case, run: $run, status: $status, secs: $secs,
tools: [$calls[] | .name], failures: ($failures | split("\n") | map(select(. != "")))}' >> "$results"
printf ' %-9s %-13s run %d %s %3ss %s\n' "$agent" "$id" "$run" "$status" "$((SECONDS - started))" \
"$(jq -r '[.[] | .name] | join(",")' <(jq -s . "$out.calls"))"
[[ -z "$failures" ]] || printf ' - %s\n' "${failures//$'\n'/$'\n - '}"
done
done
done
echo
echo "agent-smoke: pass rate per agent and case"
jq -rs 'group_by(.agent)[] | "\(.[0].agent): " + (group_by(.case) | map("\(.[0].case) \(map(select(.status=="pass")) | length)/\(length)") | join(" "))' "$results"
consistent_failures="$(jq -s '[group_by([.agent, .case])[] | select(all(.status == "fail"))] | length' "$results")"
if [[ "$consistent_failures" -gt 0 ]]; then
echo "agent-smoke: FAIL ($consistent_failures agent/case pairs failed every run)"
exit 1
fi
echo "agent-smoke: PASS"

38
tests/agents/cases.json Normal file
View file

@ -0,0 +1,38 @@
[
{
"id": "shareholders",
"prompt": "Who are the shareholders above 5% of BBCA?",
"expect_tools": [{"name": "get_shareholders", "args": {"ticker": "BBCA"}}],
"expect_text": ["(?i)dwimuria", "54[.,]9"]
},
{
"id": "quote",
"prompt": "What is the latest BBCA share price?",
"expect_tools": [{"name": "get_quotes", "args": {"symbols": ["BBCA"]}}],
"expect_text": ["(?i)bbca"]
},
{
"id": "compare",
"prompt": "Compare the fundamentals of BBCA and BBRI.",
"expect_tools": [{"name": "compare_stocks", "args": {"symbols": ["BBCA", "BBRI"]}}],
"expect_text": ["(?i)bbri"]
},
{
"id": "technical",
"prompt": "Give me a technical analysis of BBCA: trend, RSI, and MACD.",
"expect_tools": [{"name": "analyze_stock", "args": {"symbol": "BBCA"}}],
"expect_text": ["(?i)rsi"]
},
{
"id": "investor",
"prompt": "Which IDX stocks does Anthoni Salim hold above 1%?",
"expect_tools": [{"any": ["get_investor_holdings", "search_investors"]}],
"expect_text": ["(?i)bbca"]
},
{
"id": "changes",
"prompt": "Which shareholders changed between the 2026-01-30 and 2026-02-27 KSEI releases?",
"expect_tools": [{"name": "get_ownership_changes", "args": {"from": "2026-01-30", "to": "2026-02-27"}}],
"expect_text": []
}
]