Skip to content
Open
Show file tree
Hide file tree
Changes from 2 commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
14 changes: 14 additions & 0 deletions CHANGELOG.md
Original file line number Diff line number Diff line change
Expand Up @@ -7,6 +7,20 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0

## [Unreleased]

### Added

- Fleet workers now surface their deliverable. The terminal `codewhale exec`
`metadata` receipt carries `visible_final_answer_excerpt`, a bounded,
secret-redacted excerpt of the final assistant reply next to the real
`visible_final_answer_chars` count; the Runtime executor attaches it to
`Completed.summary` and, for a task with no scorer and no file artifact,
to the receipt notes instead of "no verifiable output" — a worker that
fails after writing most of a report keeps the text too. `session_capture`
now carries the raw `saved_session_id` (the `metadata` receipt stays
fingerprint-only), `FleetReceipt.saved_session_id` persists it, and the
runtime API exposes it so a client can resolve the worker's full final
reply via `GET /v1/sessions/{id}` (#5946, thanks @gaord).
Comment thread
devin-ai-integration[bot] marked this conversation as resolved.
Outdated

### Changed

- `/statusline` drives the bottom chrome again. Since the 0.9.12 shell
Expand Down
9 changes: 9 additions & 0 deletions crates/protocol/src/fleet.rs
Original file line number Diff line number Diff line change
Expand Up @@ -1126,6 +1126,12 @@ pub struct FleetReceipt {
/// existed) deserializable.
#[serde(default, skip_serializing_if = "Option::is_none")]
pub resolved_route: Option<FleetResolvedRoute>,
/// Saved exec session id holding the worker's full transcript, when the
/// worker persisted one on completion (the exec stream's
/// `session_capture.saved_session_id`). Callers resolve the final
/// assistant reply via `GET /v1/sessions/{id}`.
#[serde(default, skip_serializing_if = "Option::is_none")]
pub saved_session_id: Option<String>,
/// Effective worker authority for this task (#3211).
#[serde(default, skip_serializing_if = "Option::is_none")]
pub effective_permissions: Option<FleetEffectivePermissions>,
Expand Down Expand Up @@ -1493,6 +1499,7 @@ mod tests {
notes: None,
}),
resolved_route: None,
saved_session_id: None,
effective_permissions: None,
};
let json = serde_json::to_string(&receipt).unwrap();
Expand Down Expand Up @@ -1521,6 +1528,7 @@ mod tests {
notes: Some("manual verification required".to_string()),
}),
resolved_route: None,
saved_session_id: None,
effective_permissions: None,
};

Expand Down Expand Up @@ -1689,6 +1697,7 @@ mod tests {
model_source: Some("task.model".to_string()),
source: "resolver".to_string(),
}),
saved_session_id: None,
effective_permissions: Some(FleetEffectivePermissions {
write: true,
network: true,
Expand Down
14 changes: 14 additions & 0 deletions crates/tui/CHANGELOG.md
Original file line number Diff line number Diff line change
Expand Up @@ -7,6 +7,20 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0

## [Unreleased]

### Added

- Fleet workers now surface their deliverable. The terminal `codewhale exec`
`metadata` receipt carries `visible_final_answer_excerpt`, a bounded,
secret-redacted excerpt of the final assistant reply next to the real
`visible_final_answer_chars` count; the Runtime executor attaches it to
`Completed.summary` and, for a task with no scorer and no file artifact,
to the receipt notes instead of "no verifiable output" — a worker that
fails after writing most of a report keeps the text too. `session_capture`
now carries the raw `saved_session_id` (the `metadata` receipt stays
fingerprint-only), `FleetReceipt.saved_session_id` persists it, and the
runtime API exposes it so a client can resolve the worker's full final
reply via `GET /v1/sessions/{id}` (#5946, thanks @gaord).

### Changed

- `/statusline` drives the bottom chrome again. Since the 0.9.12 shell
Expand Down
4 changes: 4 additions & 0 deletions crates/tui/src/exec_agent.rs
Original file line number Diff line number Diff line change
Expand Up @@ -937,6 +937,7 @@ pub(crate) async fn run_exec_agent(
if let Some(id) = saved_session_id.as_ref() {
emit_exec_stream_event(&ExecStreamEvent::SessionCapture {
content: exec_stream_session_ref(id),
saved_session_id: id.clone(),
})?;
}
// Resolved output ceiling and its provenance, surfaced so a
Expand Down Expand Up @@ -989,6 +990,9 @@ pub(crate) async fn run_exec_agent(
latest_system_prompt.as_ref(),
),
visible_final_answer_chars: summary.output.chars().count(),
visible_final_answer_excerpt: exec_stream_final_answer_excerpt(
&summary.output,
),
Comment thread
devin-ai-integration[bot] marked this conversation as resolved.
resume_command: saved_session_id
.as_deref()
.map(exec_stream_resume_hint)
Expand Down
1 change: 1 addition & 0 deletions crates/tui/src/fleet/alerts.rs
Original file line number Diff line number Diff line change
Expand Up @@ -715,6 +715,7 @@ mod tests {
notes: Some("regex scorer could not be compiled".to_string()),
}),
resolved_route: None,
saved_session_id: None,
effective_permissions: None,
};

Expand Down
1 change: 1 addition & 0 deletions crates/tui/src/fleet/control.rs
Original file line number Diff line number Diff line change
Expand Up @@ -753,6 +753,7 @@ mod tests {
model_source: None,
source: "resolver".to_string(),
}),
saved_session_id: None,
effective_permissions: None,
}
}
Expand Down
Loading
Loading