Repository navigation
docs(bot): unquoted placeholder values so scan examples aren't findings - #20
Conversation
Verdict matrix
Sensitive paths
Deterministic pre-checks
gateway_safetyVERDICT: APPROVE RATIONALE
api_compatAPI Compatibility ReviewSummaryThis PR is primarily a documentation/placeholder cleanup to eliminate false-positive scan findings, with two minor code changes. The diff touches no route handlers, wire-shape logic, converters, session management, retry logic, or model alias rules. Checklist evaluation
FindingsBLOCKING: (none) SHOULD_FIX:
NITS:
Test gaps
VERDICT: APPROVE The two code changes (comment clarification in docs_configVERDICT: COMMENT BLOCKING:
SHOULD_FIX:
NITS:
Files that must change to close gap (if verification fails):
FINAL_VERDICT: COMMENT Session: response_id=resp_20aiat3vDb6i9tMPjZjWiAo,resp_req_vrtx_011CeuF1x3uWeQYXquyxru63,chatcmpl-e2be49e1-4299-4df4-a7b8-6c53646004b2 (chain via previous_response_id, 30min TTL). model=[model] provider=[provider]. |
… import - README + development: Node 18+ -> >=22.19 to match engines floor - configuration.md: canonical tool-switch header, MANAGE_BACKEND row with dual-entry note, sync inline example with config.json.example - .env.example: document OPENCODE_SERVER_URL override - proxy.ts: correct shared-verifier comment (two instances, copied keys) - grounding.ts: reuse shared asRecord from utils/guards (no behavior change)
Last dry_run had 2 nit findings left, both our own quoted '...' placeholders in docs examples. Rewrite as unquoted your-*-here placeholders. Live scan now reports 0 findings; next real run will only file true signal.