ddmitry/adversarial-review
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix: permissions, temp file prefix, session ID extraction

Browse Source 
- Fix recommended permissions: add missing Write(/tmp/codex-prompt-*),
  remove overbroad rm rule (cleanup is best-effort)
- Rename claude-plan-* → codex-plan-* so all temp files share codex-* prefix
- Extract session ID via Read tool instead of grep (no extra permission needed)
- Add UUID format spec for session ID validation

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
ddadmin
2026-04-06 23:23:07 +03:00
co-authored by Claude Opus 4.6
parent 834a74c798
commit 05fe11a4df
2 changed files with 7 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files
SKILL.md
+5 -9
View File
@@ -75,7 +75,7 @@ Save the result as `BASE_BRANCH` — used in `git diff ${BASE_BRANCH}...HEAD` be
**Plan review:**
- If the plan already exists as a file (in `project/`, plan file from Plan Mode, memory, or somewhere in the repo) — use the path directly. Do NOT copy. In Claude Code Plan Mode the plan is always a file.
- If the plan is only in the conversation context (outside Plan Mode) — write via **Write tool** to `/tmp/claude-plan-${REVIEW_ID}.md`.
- If the plan is only in the conversation context (outside Plan Mode) — write via **Write tool** to `/tmp/codex-plan-${REVIEW_ID}.md`.
- **Always print the plan file path for the user** so they can open it in their IDE:
`Plan for review: <file-path>`
@@ -314,13 +314,9 @@ timeout 600 codex exec \
- If exit code = 124 (timeout) — inform the user and offer to retry.
- stderr is redirected to a temp file for session ID capture and error diagnostics.
**After launch:** extract the session ID from the stderr file:
**After launch:** extract the session ID from the stderr file using **Read tool** on `/tmp/codex-stderr-${REVIEW_ID}.txt`. Find the line `session id: <uuid>` (format: `xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx`) and extract the UUID value.
```bash
grep -o 'session id: [a-f0-9-]*' /tmp/codex-stderr-${REVIEW_ID}.txt | head -1 | sed 's/session id: //'
```
Save the result as `CODEX_SESSION_ID` — needed for `resume` in subsequent rounds. If grep returns empty — session ID not available, resume will not work (fallback to fresh exec).
Save the result as `CODEX_SESSION_ID` — needed for `resume` in subsequent rounds. If the session ID is not found or does not match UUID format — resume will not work (fallback to fresh exec).
**Notes:**
- Default model: `gpt-5.4` with `model_reasoning_effort=high`. User can override via arguments.
@@ -445,8 +441,8 @@ Return to **Step 5**.
**Outside Plan Mode:**
```bash
rm -f /tmp/claude-plan-${REVIEW_ID}.md /tmp/codex-review-${REVIEW_ID}.md \
/tmp/codex-prompt-${REVIEW_ID}.md /tmp/codex-stderr-${REVIEW_ID}.txt
rm -f /tmp/codex-prompt-${REVIEW_ID}.md /tmp/codex-review-${REVIEW_ID}.md \
/tmp/codex-stderr-${REVIEW_ID}.txt /tmp/codex-plan-${REVIEW_ID}.md
```
If the user declined rm — continue without error.