core: persist initial context window metadata (#29519)

## Why

PR #29494 made context-window IDs visible to the model by wrapping the
token-budget window payload in `<context_window>`, but rollout JSONL
consumers still could not see the initial window identity by tailing the
session file. Compacted rollout items carry window IDs only after
compaction has happened, so a session with no compaction had no durable
JSONL record for window 0.

This change gives tailing consumers a stable initial-window record at
session creation time.

## What Changed

- Added `session_meta.context_window.window_id` for the initial
context-window identity.
- `CreateThreadParams` now requires `initial_window_id: String`, so
thread-store callers cannot accidentally create new threads without
window-0 metadata.
- Live thread creation derives the persisted initial window ID from the
same `AutoCompactWindowIds` used to initialize `SessionState`, keeping
runtime state and JSONL metadata aligned.
- Rollout reconstruction uses `session_meta.context_window.window_id` as
the initial-window fallback and derives `window_number = 0`,
`first_window_id = window_id`, and `previous_window_id = None`
internally.
- Fork reconstruction intentionally uses the same rollout reconstruction
path; consumers that need to distinguish copied initial-window metadata
can use the rollout `thread_id`.
- Legacy compactions without `window_number` still use compaction-count
fallback accounting instead of being reset to window 0 by the
initial-window fallback.
- Compacted rollout metadata still takes precedence once compaction
records exist, preserving the richer chain fields there.

## JSONL Shape

Real rollout JSONL is one object per line. This example is expanded for
readability, but shows the new initial `session_meta.context_window`
record followed by the existing compacted rollout item shape that also
carries window IDs:

```jsonl
{
  "timestamp": "2026-06-22T12:00:00.000Z",
  "type": "session_meta",
  "payload": {
    "session_id": "<THREAD_ID>",
    "id": "<THREAD_ID>",
    "timestamp": "2026-06-22T12:00:00.000Z",
    "cwd": "/repo",
    "originator": "codex",
    "cli_version": "0.0.0",
    "source": "cli",
    "model_provider": "<MODEL_PROVIDER>",
    "context_window": {
      "window_id": "<INITIAL_WINDOW_ID>"
    }
  }
}
...
{
  "timestamp": "2026-06-22T12:34:56.000Z",
  "type": "compacted",
  "payload": {
    "message": "<COMPACTION_SUMMARY>",
    "replacement_history": [
      "..."
    ],
    "window_number": 1,
    "first_window_id": "<INITIAL_WINDOW_ID>",
    "previous_window_id": "<INITIAL_WINDOW_ID>",
    "window_id": "<NEXT_WINDOW_ID>"
  }
}
```

The nested `context_window` object is intentional: it gives rollout
consumers a stable namespace for context-window metadata while only
writing the non-derivable initial `window_id`. For the initial window,
`window_number`, `first_window_id`, and `previous_window_id` are derived
internally instead of being written to the rollout.

## Verification

- `just test -p codex-protocol`
- `just test -p codex-rollout
recorder_materializes_on_flush_with_pending_items`
- `just test -p codex-core reconstruct_history`
- `just test -p codex-core
record_initial_history_reconstructs_forked_transcript`
- `just test -p codex-thread-store`
- `just test -p codex-state`
- `just test -p codex-app-server
thread_read_returns_summary_without_turns`
- `just test -p codex-rollout persistence_metrics`
This commit is contained in:
Michael Bolin
2026-06-23 21:50:50 +00:00
committed by GitHub
parent c26f961b85
commit 01f89c8c59
30 changed files with 254 additions and 13 deletions
@@ -474,6 +474,7 @@ fn write_rollout(path: &std::path::Path, thread_id: ThreadId, message: &str) ->
dynamic_tools: None,
memory_mode: None,
multi_agent_version: None,
context_window: None,
},
git: None,
};
+3
View File
@@ -51,6 +51,7 @@ async fn extract_metadata_from_rollout_uses_session_meta() {
dynamic_tools: None,
memory_mode: None,
multi_agent_version: None,
context_window: None,
};
let session_meta_line = SessionMetaLine {
meta: session_meta,
@@ -107,6 +108,7 @@ async fn extract_metadata_from_rollout_returns_latest_memory_mode() {
dynamic_tools: None,
memory_mode: None,
multi_agent_version: None,
context_window: None,
};
let polluted_meta = SessionMeta {
memory_mode: Some("polluted".to_string()),
@@ -375,6 +377,7 @@ fn write_rollout_in_sessions_with_cwd(
dynamic_tools: None,
memory_mode: None,
multi_agent_version: None,
context_window: None,
};
let session_meta_line = SessionMetaLine {
meta: session_meta,
+16
View File
@@ -57,6 +57,7 @@ use codex_protocol::protocol::MultiAgentVersion;
use codex_protocol::protocol::ResumedHistory;
use codex_protocol::protocol::RolloutItem;
use codex_protocol::protocol::RolloutLine;
use codex_protocol::protocol::SessionContextWindow;
use codex_protocol::protocol::SessionMeta;
use codex_protocol::protocol::SessionMetaLine;
use codex_protocol::protocol::SessionSource;
@@ -91,6 +92,7 @@ pub enum RolloutRecorderParams {
base_instructions: BaseInstructions,
dynamic_tools: Vec<DynamicToolSpec>,
multi_agent_version: Option<MultiAgentVersion>,
initial_window_id: Option<String>,
},
Resume {
path: PathBuf,
@@ -178,6 +180,7 @@ impl RolloutRecorderParams {
base_instructions,
dynamic_tools,
multi_agent_version: None,
initial_window_id: None,
}
}
@@ -202,6 +205,17 @@ impl RolloutRecorderParams {
self
}
pub fn with_initial_window_id(mut self, initial_window_id: String) -> Self {
if let Self::Create {
initial_window_id: window_id,
..
} = &mut self
{
*window_id = Some(initial_window_id);
}
self
}
pub fn resume(path: PathBuf) -> Self {
Self::Resume { path }
}
@@ -715,6 +729,7 @@ impl RolloutRecorder {
base_instructions,
dynamic_tools,
multi_agent_version,
initial_window_id,
} => {
let log_file_info = precompute_log_file_info(config, conversation_id)?;
let path = log_file_info.path.clone();
@@ -752,6 +767,7 @@ impl RolloutRecorder {
},
memory_mode: (!config.generate_memories()).then_some("disabled".to_string()),
multi_agent_version,
context_window: initial_window_id.map(SessionContextWindow::new),
};
(None, Some(log_file_info), path, Some(session_meta))
+11 -1
View File
@@ -103,6 +103,7 @@ async fn state_db_init_backfills_before_returning() -> anyhow::Result<()> {
dynamic_tools: None,
memory_mode: None,
multi_agent_version: None,
context_window: None,
},
git: None,
};
@@ -375,6 +376,7 @@ async fn recorder_materializes_on_flush_with_pending_items() -> std::io::Result<
let config = test_config(home.path());
let session_id = SessionId::default();
let thread_id = ThreadId::new();
let initial_window_id = Uuid::now_v7().to_string();
let recorder = RolloutRecorder::new(
&config,
RolloutRecorderParams::new(
@@ -386,7 +388,8 @@ async fn recorder_materializes_on_flush_with_pending_items() -> std::io::Result<
BaseInstructions::default(),
Vec::new(),
)
.with_session_id(session_id),
.with_session_id(session_id)
.with_initial_window_id(initial_window_id.clone()),
)
.await?;
@@ -437,6 +440,13 @@ async fn recorder_materializes_on_flush_with_pending_items() -> std::io::Result<
panic!("expected session metadata in rollout");
};
assert_eq!(session_meta.meta.session_id, session_id);
assert_eq!(
session_meta
.meta
.context_window
.map(|window| window.window_id),
Some(initial_window_id)
);
let buffered_idx = text
.find("buffered-event")
.expect("buffered event in rollout");
@@ -43,6 +43,7 @@ fn write_rollout_with_metadata(path: &Path, thread_id: ThreadId) -> std::io::Res
dynamic_tools: None,
memory_mode: None,
multi_agent_version: None,
context_window: None,
},
git: None,
}),
+1
View File
@@ -176,6 +176,7 @@ fn write_rollout_with_user_message(
dynamic_tools: None,
memory_mode: None,
multi_agent_version: None,
context_window: None,
},
git: None,
}),
+1
View File
@@ -1290,6 +1290,7 @@ async fn test_updated_at_uses_file_mtime() -> Result<()> {
dynamic_tools: None,
memory_mode: None,
multi_agent_version: None,
context_window: None,
},
git: None,
}),