You cannot select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
26 lines
1.5 KiB
Markdown
26 lines
1.5 KiB
Markdown
---
|
|
# claudbg-80fb
|
|
title: 'TUI sub-agent transcript always empty: sessionId serde rename missing'
|
|
status: completed
|
|
type: bug
|
|
priority: normal
|
|
created_at: 2026-03-31T22:32:07Z
|
|
updated_at: 2026-03-31T22:33:04Z
|
|
---
|
|
|
|
When selecting a sub-agent in the TUI transcript screen, the transcript is always empty.
|
|
|
|
Root cause: RawEntry.session_id has no #[serde(rename = "sessionId")] attribute. The JSONL field is camelCase ("sessionId") but serde only matches literal field names, so entry.session_id is always None after deserialization.
|
|
|
|
This causes read_session_id_from_first_line (discovery.rs) to always return None, so AgentRef.session_id defaults to empty string.
|
|
|
|
In load_transcript_for_agent, the find predicate uses s.session_id.starts_with(parent_session_id) — and in Rust, any_string.starts_with("") is always true. So it matches the first session in filesystem order (wrong session), fails to find the agent there, and returns with no entries loaded.
|
|
|
|
Fix: Add #[serde(rename = "sessionId")] to RawEntry.session_id in models/session.rs.
|
|
|
|
## Summary of Changes
|
|
|
|
Added `#[serde(rename = "sessionId")]` to `RawEntry.session_id` in `src/models/session.rs`.
|
|
|
|
This causes `read_session_id_from_first_line` to correctly parse the parent session UUID from agent JSONL files, so `AgentRef.session_id` is populated with the real UUID instead of defaulting to empty string. The TUI `load_transcript_for_agent` function then correctly locates the parent session and loads the agent transcript.
|