Kind
scout
Goal
Produce evidence for whether Codex Desktop delivers the same hooks and state boundaries as Codex CLI.
Non-goals
- Claiming support based only on CLI documentation.
- Implementing undocumented desktop workarounds.
- Blocking the CLI adapter on unverified desktop behavior.
Invariants
- Each tested event has a reproducible observation.
- Desktop findings are recorded separately from CLI findings.
- A failed probe results in conditional or advisory support, not silent promotion.
Behavioral contract / States
Run a desktop smoke matrix for session start, prompt submit, pre-tool, post-tool, compaction, stop, session end, multiple worktrees, and trust/config loading. Record observed input/output and enforcement result.
Scope surface
Desktop smoke fixtures, probe scripts, evidence notes, and compatibility decision.
Acceptance
- Execute the complete matrix against the installed Codex Desktop version.
- Record pass/fail/unknown for every event and the exact version tested.
- Add or update a test that fails if a claimed full-support event is not observed.
- The issue ends with one of: full adapter support, conditional support with documented limits, or advisory-only support.
Split triggers
- Desktop and CLI need separate adapters.
- Desktop has no deterministic pre-tool enforcement.
Dependencies / Parent
Blocked by: #7
Parent Epic: #15
Kind
scout
Goal
Produce evidence for whether Codex Desktop delivers the same hooks and state boundaries as Codex CLI.
Non-goals
Invariants
Behavioral contract / States
Run a desktop smoke matrix for session start, prompt submit, pre-tool, post-tool, compaction, stop, session end, multiple worktrees, and trust/config loading. Record observed input/output and enforcement result.
Scope surface
Desktop smoke fixtures, probe scripts, evidence notes, and compatibility decision.
Acceptance
Split triggers
Dependencies / Parent
Blocked by: #7
Parent Epic: #15