| name | history |
| description | Use when the user asks about review verdicts over time — "what's been getting rejected", "rejection rate this week", "show recent reviews", "history of verdicts", "which rule fails most often", "show suppressed reviews", or filtering reviews by rule / file / date / verdict / tier / severity. Zero LLM cost — pure read of stored review records. Skip when the user wants a FRESH verdict (use autoreview:review) or when no `.autoreview/` exists (use autoreview:setup first). |
AutoReview History
Cross-platform. Snippets below use bash-style env-var syntax (${CLAUDE_PLUGIN_ROOT}). Claude Code's Bash tool runs Git Bash on Windows so these work as-is; on native PowerShell substitute $env:CLAUDE_PLUGIN_ROOT, on cmd use %CLAUDE_PLUGIN_ROOT%. Plugin requires Node ≥22 — only assumed binary.
Query the verdict log written by every validate run. Zero LLM cost — purely a read of stored records.
Default — no args
node ${CLAUDE_PLUGIN_ROOT}/scripts/bin/history.mjs
Prints total record count, verdicts grouped by type, by rule, and the 10 most recent records. Good first invocation when the user says "what's been getting rejected lately?".
Common filters
--rule <id> — only records for one rule (e.g. --rule api/validate-input).--verdict pass|fail|error|suppressed — narrow to a single outcome.--file '<glob>' — glob filter on file paths (e.g. --file 'src/api/**').--since YYYY-MM-DD --until YYYY-MM-DD — date range. Either end is optional.--tier <name> — filter to records where the rule ran on a specific tier (e.g. --tier critical). Useful for auditing what the heavy reviewer is catching.--severity <error|warning> — filter to records where the rule's severity matches. Useful for --severity warning to review non-blocking findings separately.
Combine freely: --rule auth/no-direct-db --verdict fail --since 2026-04-01 --tier critical.
Output formats
--format table (default) — aggregates + the 10 most recent.--format json — single JSON object: { total, by_verdict, by_rule, records }. Good for "give me the totals as JSON".--format jsonl — one record per line, raw. Pipe through jq (POSIX) / parse with node -e "for await (const line of process.stdin) ..." (cross-platform) for custom queries.
Record fields
Each record includes: rule, verdict, reason, provider, model, mode, duration_ms, usage, actor, host, commit_sha, plus two fields added in the tier redesign:
tier — the tier that handled the rule (effective post-overlay).severity — the rule's severity at run time (effective post-overlay).
Storage layout
History lives at .autoreview/.history/<YYYY-MM-DD>.jsonl (one file per day). Long reason fields spill to sidecar files referenced by reason_sidecar in the record — read those with the Read tool when the inline reason is truncated.
Reporting
Quote the table or relevant records back to the user. For "rejection rate" questions, compute from by_verdict totals. For "which rule fails most" questions, sort by_rule by fail count.
