Clawtrix Dev Intel
Finds the best ClawHub skills for developer-tooling agents. Personalized to your tech stack and mission — not a generic popularity list.
Quick Reference
| Task | Action |
|---|---|
| New agent onboarding | Run full discovery for the agent's stated tech stack |
| Weekly skill update | Run Step 1 only — check for new releases on watched slugs |
| Specific capability gap | Run Step 2 with targeted search queries |
| Stack audit | Run full sequence, output to memory/ |
Discovery Run Sequence
Step 1 — Read Agent Mission
Read the agent's SOUL.md (or equivalent). Extract:
- Primary language/stack (e.g., TypeScript, Python, Go, Rails)
- Dev workflows the agent runs (e.g., "reviews PRs", "runs tests", "writes CI pipelines")
- Current installed skills (to avoid re-recommending what's already there)
- Any explicit gaps ("I wish I could...")
Step 2 — Search ClawHub for Dev Tooling Skills
Query the ClawHub search API (clawhub.ai/api/v1/search) for each of these categories, substituting the agent's actual stack:
- CI/CD skills — query:
ci cd - Testing and coverage — query:
testing coverage - Code review automation — query:
code review - Stack-specific — query:
{agent's primary stack} developer(e.g.,typescript developer,python developer) - Git workflow — query:
git workflow
For each result, record: slug, displayName, score, updatedAt. Deduplicate across queries.
Step 3 — Score Each Candidate
Apply the Clawtrix scoring matrix (from clawtrix-scoring-engine):
| Dimension | Max | How to measure |
|---|---|---|
| Mission relevance | 3 | Does this directly support the agent's dev workflow? 3=core, 2=adjacent, 1=tangential |
| Gap fill | 2 | Does the agent lack this capability today? 2=yes, 1=partial, 0=no |
| Community signal | 1 | installs > 1,000 = +1 |
| Recency | 1 | Updated in last 30 days = +1 |
| Trust | 1 | No security flags, legitimate publisher = +1 |
Step 4 — Apply Dev-Specific Filters
Before recommending, verify:
- Skill is compatible with the agent's primary language/framework
- No security flags (run against
clawtrix-security-auditpattern list) - Publisher has a credible track record (> 2 other published skills)
- Not already installed by this agent
Step 5 — Output Top 3
Never recommend more than 3. Write to memory/reports/dev-intel-YYYY-MM-DD.md:
# Dev Intel — YYYY-MM-DD
## Agent: [name]
## Stack: [languages/frameworks]
## Skills audited: N candidates
## Top 3 Recommendations
**1. [author/slug]** (score: N/8)
- What: [one sentence]
- Why for this agent: [one sentence tied to SOUL.md]
- Install: `clawhub install [slug]`
**2. ...**
**3. ...**
## Skipped (and why)
| Slug | Reason |
|------|--------|
| ... | Low mission relevance / security flag / already installed |
Watched Skills
These are the highest-signal dev-tooling skills on ClawHub based on first intelligence run:
| Slug | What it does | Why it matters |
|---|---|---|
pskoett/self-improving-agent |
Captures learnings, errors, corrections | Continuous improvement loop for coding agents |
agent-team-orchestration |
Multi-agent task routing and handoffs | Essential for agents coordinating with Paperclip or other agents |
security-audit-toolkit |
OWASP checks, codebase vulnerability scanning | Any agent touching production code needs this layer |
When to Use This for n8n Teams
For dev agents at companies already running n8n automation:
- ClawHub skills for n8n workflow conversion exist and are a natural fit
- These teams already automate workflows — adding skill-based AI extends that investment seamlessly
Run Step 2 with n8n workflow as a search query to find current conversion options in ClawHub.
Upgrade Note — Clawtrix Pro
This skill surfaces dev-tooling recommendations on demand. Clawtrix Pro adds:
- Proactive alerts when a high-signal dev skill ships or updates
- Cross-agent comparison ("your CTO agent has X but your dev agent doesn't")
- Weekly dev stack briefing with install/update diffs
Version History
v0.1.0 — Initial release. Stack-personalized discovery, 5-query search sequence, scoring matrix integration, n8n angle included. v0.1.2 — 2026-04-02 — Replaced bash curl blocks in Step 2 with descriptive search instructions to resolve scanner flag. v0.1.1 — Cleaned up internal research notes from n8n and watched-skills sections; now fully customer-facing.