artifact-integrator

Deep integration analysis for newly created artifacts

Zur Installation springen

Repository: oimiragieo/agent-studio
Letzte Quellaktivität: 21. April 2026 um 23:45
Erkannte Sprache von SKILL.md: Englisch
Sterne: 36
Forks: 6

Installationsoptionen

Standardmäßig ist der Prompt ausgewählt, der zuerst die Quelle prüft. Sie können zu einem direkten Befehl wechseln oder eine lokale Kopie herunterladen.

Quelldateien prüfen

Lesen Sie SKILL.md und alle von SkillsMP angezeigten Begleitdateien, bevor Sie sich für eine Installation entscheiden.

Zurück zur Suche Zurück zur Installation

name	artifact-integrator
description	Deep integration analysis for newly created artifacts
version	1.1.0
category	workflow
agents	["architect","planner","developer"]
tools	["Read","Write","Edit","TaskCreate","TaskList","TaskUpdate","Bash"]
verified	true
lastVerifiedAt	"2026-02-22T00:00:00.000Z"
source	builtin
trust_score	100
provenance_sha	6286c3c564ac5576

Artifact Integrator

Purpose

Analyze newly created or modified artifacts for integration gaps and propose follow-up tasks to ensure full ecosystem integration.

When to Use

After any artifact creator skill completes
When Router Step 0.5 detects unprocessed integration queue entries
When manually checking integration health
When reviewing artifacts for completeness

Protocol

Step 1: Read Integration Queue

Read .claude/context/runtime/integration-queue.jsonl Filter for entries with "processed": false If no unprocessed entries, check if artifactId argument was provided for direct analysis.

Step 2: Analyze Each Artifact

For each unprocessed entry (or the directly specified artifact):

Load the artifact graph: require('.claude/lib/workflow/artifact-graph.cjs')
Load impact analyzer: require('.claude/lib/workflow/integration-impact.cjs')
Call analyzeImpact({ artifactId, changeType, graphPath })
Review the missingIntegrations and proposedTasks

Step 3: Generate Integration Plan

For each artifact with missing integrations:

Must-Have (P1) — Create tasks immediately:

Missing catalog entry → TaskCreate: "Add {name} to {catalog}"
Missing agent assignment → TaskCreate: "Assign {name} to relevant agent"
Missing routing entry → TaskCreate: "Update routing for {name}"
Missing hook registration → TaskCreate: "Register {name} in settings.json"

Should-Have (P2) — Create tasks with lower priority:

Missing documentation → TaskCreate: "Document {name} in {doc}"
Missing enforcement hook → TaskCreate: "Create enforcement for {name}"
Missing tests → TaskCreate: "Write tests for {name}"

Nice-to-Have — Note but don't create tasks:

Missing templates, optional docs

Step 3.1: Companion Matrix Analysis (Interwoven Creator Ecosystem)

For each artifact analyzed, run companion matrix validation to identify missing ecosystem companions:

Analysis:

Use companion-check.cjs from .claude/lib/creators/companion-check.cjs
Call to get companion matrix

checkCompanions(artifactType, artifactName)

### Companion Matrix Analysis: skill:rate-limiter

**Required Companions (MISSING):**

- [ ] hook:rate-limit-validator (validates rate limit headers) — **TASK CREATED** (P1)
- [ ] schema:rate-limit-config (validates configuration) — **TASK CREATED** (P1)

**Recommended Companions (MISSING):**

- [ ] template:rate-limit-pattern (template for rate limiting implementations) — **TASK CREATED** (P2)

**Optional Companions (MISSING):**

- [ ] workflow:rate-limit-setup (onboarding workflow) — noted for future consideration

**Safety Check:**

- Auto-spawned tasks: 3 (within limit of 3 per artifact)
- Total companion tasks this run: 8 (within limit of 10)

## Backward Propagation Analysis

**Pattern**: <validated pattern description>
**Instances Found**: <actual count> (claimed: <original count>)
**Proposed Artifact**: <type>:<name>
**Justification**: <why centralization is beneficial>
**Priority**: P1 | P2
**Next Steps**: Queued for creator skill invocation

{
  "artifactId": "hook:jwt-validation",
  "changeType": "backward-propagation",
  "timestamp": "2026-02-08T10:30:00Z",
  "processed": false,
  "source": "code-reviewer",
  "pattern": "Manual JWT validation logic duplicated in 4 route handlers",
  "affectedFiles": [
    "routes/auth.ts",
    "routes/api.ts",
    "routes/admin.ts",
    "routes/user.ts"
  ],
  "validatedInstances": 4,
  "estimatedLOCReduction": 120,
  "priority": "P1",
  "proposedArtifact": {
    "type": "hook",
    "name": "jwt-validation",
    "rationale": "Centralize JWT validation for consistent security enforcement"
  }
}

## Integration Analysis Report

Processed: {count} artifacts
Tasks created: {count}
Must-have gaps: {count}
Should-have gaps: {count}

### Details
[artifact-by-artifact breakdown]

Type	Must-Have	Should-Have
Skill	Catalog + agent assignment	Hook, workflow ref
Agent	Registry + routing keywords	Skills, model config
Hook	settings.json registration	Docs entry
Workflow	Registry + agent mapping	Docs entry
Template	Catalog entry	Consumer ref
Schema	Catalog entry	Consumer wiring

Skill({ skill: 'artifact-integrator' });
// Processes queue, creates tasks, updates graph

Skill({ skill: 'artifact-integrator', args: 'skill:rate-limiter' });
// Analyzes specific artifact

{"artifactId":"skill:rate-limiter","changeType":"created","timestamp":"2026-02-07T10:30:00Z","processed":false}
{"artifactId":"agent:security-architect","changeType":"updated","timestamp":"2026-02-07T10:35:00Z","processed":false}

// Router Step 0.5 pseudocode
if (integrationQueueHasUnprocessedEntries()) {
  Task({
    task_id: 'task-1',
    subagent_type: 'developer',
    prompt: 'Invoke Skill({ skill: "artifact-integrator" })',
  });
}

Anti-Pattern	Why It Fails	Correct Approach
Skipping integration queue check	Unprocessed entries accumulate silently	Always read `integration-queue.jsonl` at Step 1
Creating >3 companion tasks per artifact	Task explosion overwhelms the agent pipeline	Enforce SEC-ICE-002: max 3 per artifact, 10 total
Queuing backward-propagation without code validation	False positives create unnecessary artifacts	Verify >= 3 instances AND >= 30 LOC reduction
Not marking queue entries processed	Entries retry indefinitely on next invocation	Set `processed: true` before reporting completion
Reporting gaps without updating artifact graph	Graph becomes stale; future analysis uses wrong data	Always update graph edges after gap analysis
Running integration on unverified artifact IDs	Invalid IDs cause silent failures	Verify artifact file exists before analysis