From silver-bullet
Orchestrates artifact review by dispatching to specialized reviewer skills, tracking state, and writing a REVIEW-ROUNDS.md audit trail.
How this skill is triggered — by the user, by Claude, or both
Slash command
/silver-bullet:artifact-reviewer <artifact-path> [--reviewer <reviewer-skill-name>]<artifact-path> [--reviewer <reviewer-skill-name>]The summary Claude sees in its skill listing — used to decide when to auto-load this skill
Orchestrator skill for artifact review. Accepts an artifact path and optional reviewer name, dispatches to the appropriate reviewer skill, runs the 2-consecutive-pass review loop, records per-artifact state, and writes the REVIEW-ROUNDS.md audit trail.
Orchestrator skill for artifact review. Accepts an artifact path and optional reviewer name, dispatches to the appropriate reviewer skill, runs the 2-consecutive-pass review loop, records per-artifact state, and writes the REVIEW-ROUNDS.md audit trail.
/artifact-reviewer <artifact-path> [--reviewer <reviewer-skill-name>]
If --reviewer is omitted, the reviewer is auto-detected from the artifact filename using the mapping table below.
| Artifact Pattern | Reviewer Skill | Notes |
|---|---|---|
| *-PLAN.md | review-plan | SB plan reviewer |
| Code changes | silver-review | SB code review workflow |
| VERIFICATION.md | review-verification | SB verification artifact reviewer |
| Security findings | security | Existing SB reviewer |
| SPEC.md | review-spec | New SB reviewer |
| DESIGN.md | review-design | New SB reviewer |
| REQUIREMENTS.md | review-requirements | New SB reviewer |
| ROADMAP.md | review-roadmap | New SB reviewer |
| CONTEXT.md | review-context | New SB reviewer |
| RESEARCH.md | review-research | New SB reviewer |
| INGESTION_MANIFEST.md | review-ingestion-manifest | New SB reviewer |
| UAT.md | review-uat | New SB reviewer |
| Cross-artifact set | review-cross-artifact | New SB reviewer -- multi-artifact |
.planning/config.json and run the review loop (defined in rules/review-loop.md) with the resolved depth
4.5. After each review round, emit metrics to .planning/review-analytics.jsonl (per rules/review-loop.md Section 4).planning/review-analytics.jsonl exceeds 1000 lines, archive to .planning/archive/review-analytics-{date}.jsonl before next appendReview depth is configured per artifact type in .planning/config.json:
{
"review_depth": {
"review-spec": "deep",
"review-roadmap": "quick",
"review-plan": "standard"
}
}
| Depth | QC Checks | Required Clean Passes | Use When |
|---|---|---|---|
| deep | Full (all QC) | 2 consecutive | High-stakes artifacts (SPEC, REQUIREMENTS) |
| standard | Full (all QC) | 1 | Default — good balance of rigor and speed |
| quick | Structural only | 1 | Speed-critical reviews (CONTEXT, RESEARCH) |
review_depth is absent from config.json: all artifacts use standardreview_depth exists but has no entry for a reviewer: that reviewer uses standardreview_depth: {} is equivalent to "all standard"Every review round emits a structured metric to .planning/review-analytics.jsonl (JSON Lines format, one object per line). Each record captures:
artifact_path, artifact_type, reviewer — what was reviewedround, finding_count, status — round outcomedepth, check_mode — review configurationduration_seconds, timestamp — timing dataWhen the analytics file exceeds 1000 lines, it is archived to .planning/archive/review-analytics-{date}.jsonl and a fresh file is started.
Run silver:review-stats to produce summary reports from the analytics data.
All reviewers implementing this framework MUST load:
@skills/artifact-reviewer/rules/reviewer-interface.md — interface contract (input/output shape all reviewers must implement)@skills/artifact-reviewer/rules/review-loop.md — loop mechanism, per-artifact state tracking, and audit trail formatnpx claudepluginhub alo-exp/silver-bullet --plugin silver-bulletGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.