@team-agent/installer 0.5.45 → 0.5.47

package/crates/team-agent/src/provider/session/capture.rs

@@ -8,6 +8,7 @@ use crate::provider::{
     CaptureSessionContext, CapturedSession, CapturedSessionCandidate, Provider, ProviderAdapter,
     ProviderError, SessionId,
 };
+use crate::state::identity_keys::SessionAttributionKey;
 
 pub const SESSION_CAPTURE_CONVERGENCE_DEADLINE_MS: u64 = 12_000;
 pub const SESSION_CAPTURE_CONVERGENCE_POLL_MS: u64 = 250;
@@ -638,6 +639,7 @@ pub fn incomplete_interacted_resumable_agent_ids(state: &Value) -> Vec<String> {
 struct PendingSessionCapture {
     agent_id: String,
     provider: Provider,
+    team_key: String,
     context: CaptureSessionContext,
 }
 
@@ -673,6 +675,13 @@ where
     Some(PendingSessionCapture {
         agent_id: agent_id.to_string(),
         provider,
+        team_key: agent
+            .get("owner_team_id")
+            .or_else(|| agent.get("team_key"))
+            .and_then(Value::as_str)
+            .filter(|s| !s.is_empty())
+            .unwrap_or("current")
+            .to_string(),
         context: CaptureSessionContext {
             agent_id: agent_id.to_string(),
             spawn_cwd: PathBuf::from(spawn_cwd),
@@ -883,14 +892,14 @@ fn allocate_session_candidates(
                     .as_ref()
                     .is_some_and(|sid| sid.as_str() == expected.as_str())
             })
-            .filter(|candidate| !candidate_keys_collide(candidate, claimed))
+            .filter(|candidate| !candidate_keys_collide(item, candidate, claimed))
             .collect();
         // Uniqueness requirement: only assign when the expected id maps to
         // exactly one available candidate. Multiple matches or a colliding
         // single match leave the agent for the ambiguity path below.
         if exact_matches.len() == 1 {
             let candidate = exact_matches[0].clone();
-            claimed.extend(captured_provider_session_keys(&candidate.captured));
+            claimed.extend(captured_provider_session_keys(item, &candidate.captured));
             assignments.insert(item.agent_id.clone(), candidate);
         }
     }
@@ -903,8 +912,9 @@ fn allocate_session_candidates(
             candidates_by_agent.get(&item.agent_id),
             claimed,
             CandidateMatchKind::PositiveAgentId,
+            item,
         ) {
-            claimed.extend(captured_provider_session_keys(&candidate.captured));
+            claimed.extend(captured_provider_session_keys(item, &candidate.captured));
             assignments.insert(item.agent_id.clone(), candidate);
         }
     }
@@ -916,8 +926,9 @@ fn allocate_session_candidates(
             candidates_by_agent.get(&item.agent_id),
             claimed,
             CandidateMatchKind::PathAgentId,
+            item,
         ) {
-            claimed.extend(captured_provider_session_keys(&candidate.captured));
+            claimed.extend(captured_provider_session_keys(item, &candidate.captured));
             assignments.insert(item.agent_id.clone(), candidate);
         }
     }
@@ -953,9 +964,10 @@ fn allocate_session_candidates(
             candidates_by_agent.get(&item.agent_id),
             claimed,
             CandidateMatchKind::Any,
+            item,
         ) {
             Some(candidate) => {
-                claimed.extend(captured_provider_session_keys(&candidate.captured));
+                claimed.extend(captured_provider_session_keys(item, &candidate.captured));
                 assignments.insert(item.agent_id.clone(), candidate);
             }
             None => {
@@ -995,11 +1007,14 @@ fn allocate_global_one_to_one(
         let Some(agent_candidates) = candidates_by_agent.get(agent_id) else {
             return;
         };
+        let Some(owner) = pending.iter().find(|item| item.agent_id == *agent_id) else {
+            continue;
+        };
         for candidate in agent_candidates {
-            if candidate_keys_collide(candidate, claimed) {
+            if candidate_keys_collide(owner, candidate, claimed) {
                 continue;
             }
-            let key = candidate_key(candidate);
+            let key = candidate_key(owner, candidate);
             if key.is_empty() {
                 continue;
             }
@@ -1010,7 +1025,9 @@ fn allocate_global_one_to_one(
         return;
     }
     for (agent_id, candidate) in remaining_agents.into_iter().zip(candidates.into_values()) {
-        claimed.extend(captured_provider_session_keys(&candidate.captured));
+        if let Some(item) = pending.iter().find(|item| item.agent_id == agent_id) {
+            claimed.extend(captured_provider_session_keys(item, &candidate.captured));
+        }
         assignments.insert(agent_id, candidate);
     }
 }
@@ -1035,6 +1052,7 @@ fn unique_available_candidate(
     candidates: Option<&Vec<CapturedSessionCandidate>>,
     claimed: &BTreeSet<String>,
     match_kind: CandidateMatchKind,
+    owner: &PendingSessionCapture,
 ) -> Option<CapturedSessionCandidate> {
     let matches = candidates?
         .iter()
@@ -1043,7 +1061,7 @@ fn unique_available_candidate(
             CandidateMatchKind::PathAgentId => candidate.agent_path_match,
             CandidateMatchKind::Any => true,
         })
-        .filter(|candidate| !candidate_keys_collide(candidate, claimed))
+        .filter(|candidate| !candidate_keys_collide(owner, candidate, claimed))
         .cloned()
         .collect::<Vec<_>>();
     if matches.len() == 1 {
@@ -1061,16 +1079,17 @@ enum CandidateMatchKind {
 }
 
 fn candidate_keys_collide(
+    owner: &PendingSessionCapture,
     candidate: &CapturedSessionCandidate,
     claimed: &BTreeSet<String>,
 ) -> bool {
-    captured_provider_session_keys(&candidate.captured)
+    captured_provider_session_keys(owner, &candidate.captured)
         .iter()
         .any(|key| claimed.contains(key))
 }
 
-fn candidate_key(candidate: &CapturedSessionCandidate) -> String {
-    captured_provider_session_keys(&candidate.captured)
+fn candidate_key(owner: &PendingSessionCapture, candidate: &CapturedSessionCandidate) -> String {
+    captured_provider_session_keys(owner, &candidate.captured)
         .into_iter()
         .collect::<Vec<_>>()
         .join("|")
@@ -1130,12 +1149,13 @@ fn claimed_provider_session_keys(
     pending_ids: &BTreeSet<String>,
 ) -> BTreeSet<String> {
     let mut keys = BTreeSet::new();
+    let team_key = crate::state::projection::team_state_key(state);
     // 1. Non-pending worker sessions (existing behaviour).
     for (agent_id, agent) in agents {
         if pending_ids.contains(agent_id) {
             continue;
         }
-        push_provider_session_keys(&mut keys, agent);
+        push_provider_session_keys(&mut keys, &team_key, agent_id, agent);
     }
     // 2. P0 (lane-046-capture-gap): leader anchor sessions. The leader's
     //    own provider transcript must never be attributed to a worker. Scan
@@ -1152,14 +1172,39 @@ fn claimed_provider_session_keys(
     keys
 }
 
-fn push_provider_session_keys(keys: &mut BTreeSet<String>, value: &Value) {
+fn push_provider_session_keys(
+    keys: &mut BTreeSet<String>,
+    fallback_team_key: &str,
+    fallback_agent_id: &str,
+    value: &Value,
+) {
+    let provider = value
+        .get("provider")
+        .and_then(Value::as_str)
+        .and_then(parse_provider)
+        .unwrap_or(Provider::Fake);
+    let team_key = value
+        .get("owner_team_id")
+        .or_else(|| value.get("team_key"))
+        .and_then(Value::as_str)
+        .filter(|s| !s.is_empty())
+        .unwrap_or(fallback_team_key);
+    let agent_id = value
+        .get("agent_id")
+        .and_then(Value::as_str)
+        .filter(|s| !s.is_empty())
+        .unwrap_or(fallback_agent_id);
     for field in ["session_id", "provider_session_id"] {
         if let Some(session_id) = value
             .get(field)
             .and_then(Value::as_str)
             .filter(|s| !s.is_empty())
         {
-            keys.insert(format!("session:{session_id}"));
+            if let Some(key) = SessionAttributionKey::new(provider, team_key, agent_id, session_id)
+            {
+                keys.insert(session_attribution_key_string(&key));
+            }
+            keys.insert(global_session_attribution_key_string(session_id));
         }
     }
     for field in ["rollout_path", "transcript_path"] {
@@ -1168,7 +1213,10 @@ fn push_provider_session_keys(keys: &mut BTreeSet<String>, value: &Value) {
             .and_then(Value::as_str)
             .filter(|s| !s.is_empty())
         {
-            keys.insert(format!("rollout:{path}"));
+            keys.insert(transcript_attribution_key_string(
+                provider, team_key, agent_id, path,
+            ));
+            keys.insert(global_transcript_attribution_key_string(path));
         }
     }
 }
@@ -1176,25 +1224,68 @@ fn push_provider_session_keys(keys: &mut BTreeSet<String>, value: &Value) {
 fn push_leader_provider_session_keys(keys: &mut BTreeSet<String>, scope: &Value) {
     for anchor in ["leader_receiver", "team_owner"] {
         if let Some(node) = scope.get(anchor) {
-            push_provider_session_keys(keys, node);
+            push_provider_session_keys(keys, "leader", "leader", node);
         }
     }
 }
 
-fn captured_provider_session_keys(captured: &CapturedSession) -> BTreeSet<String> {
+fn captured_provider_session_keys(
+    owner: &PendingSessionCapture,
+    captured: &CapturedSession,
+) -> BTreeSet<String> {
     let mut keys = BTreeSet::new();
+    let team_key = owner.team_key.as_str();
     if let Some(session_id) = &captured.session_id {
-        keys.insert(format!("session:{}", session_id.as_str()));
+        if let Some(key) = SessionAttributionKey::new(
+            owner.provider,
+            team_key,
+            owner.agent_id.as_str(),
+            session_id.as_str(),
+        ) {
+            keys.insert(session_attribution_key_string(&key));
+        }
+        keys.insert(global_session_attribution_key_string(session_id.as_str()));
     }
     if let Some(rollout_path) = &captured.rollout_path {
-        keys.insert(format!(
-            "rollout:{}",
-            rollout_path.as_path().to_string_lossy()
+        let path = rollout_path.as_path().to_string_lossy();
+        keys.insert(transcript_attribution_key_string(
+            owner.provider,
+            team_key,
+            owner.agent_id.as_str(),
+            &path,
         ));
+        keys.insert(global_transcript_attribution_key_string(&path));
     }
     keys
 }
 
+fn session_attribution_key_string(key: &SessionAttributionKey) -> String {
+    format!(
+        "SessionAttributionKey(provider={:?},team={},agent={},session={})",
+        key.provider(),
+        key.team_key(),
+        key.agent_id(),
+        key.session_id()
+    )
+}
+
+fn transcript_attribution_key_string(
+    provider: Provider,
+    team_key: &str,
+    agent_id: &str,
+    path: &str,
+) -> String {
+    format!("TranscriptAttributionKey(provider={provider:?},team={team_key},agent={agent_id},path={path})")
+}
+
+fn global_session_attribution_key_string(session_id: &str) -> String {
+    format!("GlobalSessionAttributionKey(session={session_id})")
+}
+
+fn global_transcript_attribution_key_string(path: &str) -> String {
+    format!("GlobalTranscriptAttributionKey(path={path})")
+}
+
 #[cfg(test)]
 pub(crate) mod test_support {
     use super::*;

package/crates/team-agent/src/provider/session_scan/codex.rs

@@ -14,10 +14,13 @@ pub(super) fn apply_spawned_at_filter(
             Some(p) => p.as_path(),
             None => return false,
         };
-        std::fs::metadata(path)
-            .and_then(|meta| meta.modified())
-            .map(|mtime| mtime >= cutoff)
-            .unwrap_or(false)
+        match codex_rollout_created_at(path) {
+            Some(created_at) => created_at >= cutoff,
+            None => std::fs::metadata(path)
+                .and_then(|meta| meta.modified())
+                .map(|mtime| mtime >= cutoff)
+                .unwrap_or(false),
+        }
     });
 }
 
@@ -27,6 +30,50 @@ pub(super) fn parse_spawned_at(raw: &str) -> Option<std::time::SystemTime> {
         .map(|dt| std::time::SystemTime::from(dt.with_timezone(&chrono::Utc)))
 }
 
+fn codex_rollout_created_at(path: &std::path::Path) -> Option<std::time::SystemTime> {
+    created_at_from_rollout_head(path).or_else(|| created_at_from_rollout_filename(path))
+}
+
+fn created_at_from_rollout_head(path: &std::path::Path) -> Option<std::time::SystemTime> {
+    let text = super::common::read_head_text(path, super::common::CAPTURE_HEAD_BYTES).ok()?;
+    super::common::parse_session_records(&text)
+        .iter()
+        .find_map(record_created_at)
+        .and_then(|raw| parse_spawned_at(&raw))
+}
+
+fn record_created_at(record: &serde_json::Value) -> Option<String> {
+    record
+        .get("created_at")
+        .and_then(serde_json::Value::as_str)
+        .or_else(|| {
+            record
+                .get("session_meta")
+                .and_then(|v| v.get("payload"))
+                .or_else(|| record.get("payload"))
+                .and_then(|v| v.get("created_at"))
+                .and_then(serde_json::Value::as_str)
+        })
+        .map(ToString::to_string)
+}
+
+fn created_at_from_rollout_filename(path: &std::path::Path) -> Option<std::time::SystemTime> {
+    let name = path.file_name()?.to_str()?;
+    let start = name.find("rollout-")? + "rollout-".len();
+    let stamp = name.get(start..start + 19)?;
+    if stamp.as_bytes().get(10).copied() != Some(b'T') {
+        return None;
+    }
+    let raw = format!(
+        "{}T{}:{}:{}+00:00",
+        &stamp[0..10],
+        &stamp[11..13],
+        &stamp[14..16],
+        &stamp[17..19]
+    );
+    parse_spawned_at(&raw)
+}
+
 #[cfg(test)]
 mod tests {
     use super::*;

package/crates/team-agent/src/redaction.rs

@@ -0,0 +1,185 @@
+use std::sync::LazyLock;
+
+use regex::{Captures, Regex};
+use serde_json::Value;
+
+const REDACTED: &str = "[REDACTED]";
+const SENSITIVE_FAMILIES: [&str; 7] = [
+    "PROXY",
+    "TOKEN",
+    "KEY",
+    "PASSWORD",
+    "AUTH",
+    "SECRET",
+    "CREDENTIAL",
+];
+const STRUCTURAL_KEYS: [&str; 13] = [
+    "active_team_key",
+    "cohort_key",
+    "dedupe_key",
+    "error_key",
+    "error_observation_key",
+    "last_check_key",
+    "last_error_observation_key",
+    "last_notified_key",
+    "last_suppressed_key",
+    "parent_team_key",
+    "runtime_team_key",
+    "team_key",
+    "team_state_key",
+];
+
+static SHELL_ASSIGNMENT: LazyLock<Regex> = LazyLock::new(|| {
+    Regex::new(
+        r#"(?i)(?P<prefix>^|[\s\[,;('"])(?P<key>[a-z_][a-z0-9_]*)=(?P<value>'[^']*'|"[^"]*"|[^\s,\]\[};)]+)"#,
+    )
+    .expect("shell assignment redaction regex")
+});
+
+static URL_USERINFO: LazyLock<Regex> = LazyLock::new(|| {
+    Regex::new(r"(?P<scheme>[A-Za-z][A-Za-z0-9+.-]*://)[^\s/?#]+@")
+        .expect("URL userinfo redaction regex")
+});
+
+pub(crate) fn redact_external_value(value: &serde_json::Value) -> serde_json::Value {
+    match value {
+        Value::Object(object) => Value::Object(
+            object
+                .iter()
+                .map(|(key, value)| {
+                    let value = if is_sensitive_env_key(key) {
+                        Value::String(REDACTED.to_string())
+                    } else {
+                        redact_external_value(value)
+                    };
+                    (key.clone(), value)
+                })
+                .collect(),
+        ),
+        Value::Array(values) => Value::Array(values.iter().map(redact_external_value).collect()),
+        Value::String(text) => Value::String(redact_external_text(text)),
+        other => other.clone(),
+    }
+}
+
+pub(crate) fn redact_external_text(text: &str) -> String {
+    let assignments = SHELL_ASSIGNMENT.replace_all(text, |captures: &Captures<'_>| {
+        let key = captures.name("key").map_or("", |value| value.as_str());
+        if !is_sensitive_env_key(key) {
+            return captures[0].to_string();
+        }
+        let prefix = captures.name("prefix").map_or("", |value| value.as_str());
+        let value = captures.name("value").map_or("", |value| value.as_str());
+        let quote = match (value.as_bytes().first(), value.as_bytes().last()) {
+            (Some(b'\''), Some(b'\'')) => "'",
+            (Some(b'"'), Some(b'"')) => "\"",
+            _ => "",
+        };
+        let unquoted = value
+            .strip_prefix(quote)
+            .and_then(|value| value.strip_suffix(quote))
+            .unwrap_or(value);
+        let has_url_userinfo = URL_USERINFO.is_match(unquoted);
+        let redacted_url = URL_USERINFO
+            .replace_all(unquoted, "${scheme}[REDACTED]@")
+            .into_owned();
+        let safe_value = if has_url_userinfo {
+            &redacted_url
+        } else {
+            REDACTED
+        };
+        format!("{prefix}{key}={quote}{safe_value}{quote}")
+    });
+    URL_USERINFO
+        .replace_all(&assignments, "${scheme}[REDACTED]@")
+        .into_owned()
+}
+
+fn is_sensitive_env_key(key: &str) -> bool {
+    if STRUCTURAL_KEYS
+        .iter()
+        .any(|structural| key.eq_ignore_ascii_case(structural))
+    {
+        return false;
+    }
+    let key = key.to_ascii_uppercase();
+    SENSITIVE_FAMILIES
+        .iter()
+        .any(|family| key == *family || key.ends_with(&format!("_{family}")))
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use serde_json::json;
+
+    #[test]
+    fn recursive_values_mask_env_families_without_erasing_wire_truth() {
+        let input = json!({
+            "HTTPS_PROXY": "proxy-secret",
+            "Copilot_Github_Token": "token-secret",
+            "nested": [{"db_password": "password-secret"}],
+            "team_key": "current",
+            "active_team_key": "current",
+            "dedupe_key": "restart:worker",
+            "auth_mode": "subscription",
+            "model_source": "role",
+            "model_stale": true,
+        });
+
+        let redacted = redact_external_value(&input);
+
+        assert_eq!(redacted["HTTPS_PROXY"], REDACTED);
+        assert_eq!(redacted["Copilot_Github_Token"], REDACTED);
+        assert_eq!(redacted["nested"][0]["db_password"], REDACTED);
+        assert_eq!(redacted["team_key"], "current");
+        assert_eq!(redacted["active_team_key"], "current");
+        assert_eq!(redacted["dedupe_key"], "restart:worker");
+        assert_eq!(redacted["auth_mode"], "subscription");
+        assert_eq!(redacted["model_source"], "role");
+        assert_eq!(redacted["model_stale"], true);
+        assert_eq!(redact_external_value(&redacted), redacted);
+    }
+
+    #[test]
+    fn text_masks_quoted_assignments_and_url_userinfo_idempotently() {
+        let input = "HTTPS_PROXY='https://user:pass@proxy.invalid:8443/path' http_proxy=other endpoint=https://user:pass@proxy.invalid:8443/path";
+        let expected = "HTTPS_PROXY='https://[REDACTED]@proxy.invalid:8443/path' http_proxy=[REDACTED] endpoint=https://[REDACTED]@proxy.invalid:8443/path";
+
+        let redacted = redact_external_text(input);
+
+        assert_eq!(redacted, expected);
+        assert_eq!(redact_external_text(&redacted), redacted);
+    }
+
+    #[test]
+    fn text_leaves_non_env_assignments_and_plain_urls_unchanged() {
+        let input = "team_key=current author=operator endpoint=https://proxy.invalid/health";
+        assert_eq!(redact_external_text(input), input);
+    }
+
+    #[test]
+    fn mixed_external_value_is_idempotent() {
+        let marker = "synthetic-redaction-unit-marker";
+        let credential_url = format!("https://demo-user:{marker}@proxy.invalid:8443/path");
+        let diagnostic = format!(
+            "subprocess exited 37: argv=[tmux, HTTPS_PROXY='{credential_url}']; endpoint={credential_url}"
+        );
+        let input = json!({
+            "HTTPS_PROXY": marker,
+            "ordinary_diagnostic": format!(
+                "HTTPS_PROXY='{credential_url}' http_proxy=\"{credential_url}\" OPENAI_API_KEY={marker} endpoint={credential_url}"
+            ),
+            "nested": [[diagnostic, credential_url]],
+            "team_key": "current",
+        });
+
+        let once = redact_external_value(&input);
+        let twice = redact_external_value(&once);
+        let text = once.to_string();
+
+        assert!(!text.contains(marker));
+        assert!(text.contains("https://[REDACTED]@proxy.invalid:8443/path"));
+        assert_eq!(twice, once);
+    }
+}