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

perf(skill): pass runner spec by path, skip Read in main

Browse Source 
Main no longer Reads references/runner.md. The Agent prompt now
passes RUNNER_SPEC_PATH as a short bootstrap instruction; the
subagent Reads the spec itself. Saves ~12K per round from main
context (Read result + inlined-spec duplication in Agent prompt).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
ddadmin
2026-04-19 18:56:17 +03:00
co-authored by Claude Opus 4.7
parent 7f184d0507
commit b9264a9001
2 changed files with 10 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/DESIGN.md
+2
View File
@@ -1453,6 +1453,8 @@ Claude Code's Agent tool dispatches a fresh subagent with its own isolated conte
The runner uses a two-channel protocol: the authoritative structured result is written to `/tmp/codex-runner-result-${REVIEW_ID}.json`, and the runner's final message is a single `RUNNER_RESULT_AT: <path>` line. Main extracts the path with a tolerant regex (markdown fences and minor wrapping do not break parsing) and reads the JSON file directly. This avoids the brittle "raw JSON in message" contract that Haiku's conversational output style would otherwise stress.
**Runner spec is passed by path, not inlined.** Main resolves `RUNNER_SPEC_PATH` (3-tier filesystem lookup in SKILL.md Step 4) and passes the absolute path to the subagent — the subagent Reads runner.md itself. Main never Reads runner.md. Rationale: inlining the full spec (~12K) into every Agent-tool prompt would re-add ~12K × rounds (up to 5) to main's context per review — 60K of avoidable overhead. Bootstrap instruction in the Agent prompt is ~400 bytes; the spec lives only in the disposable subagent context.
### §12.3 Retry budget — single owner
Retry lives in the runner alone, across ALL failure types (launch_failure, timeout, stderr-infrastructure-error). Runner retries once internally (same ATTEMPT_ID-rotation as pre-refactor's round-level retry). Main treats EVERY failure result as terminal-for-this-round: on the initial Step-4 dispatch, terminal = abort; on the Step-7 resume dispatch, terminal = route to fallback (fresh-exec, which is a new round with its own budget).