From sentinel
Scan global skills against project context and produce a tool routing report for developer review. Use when initializing a project, onboarding new tools, or when the developer says "routing", "scan tools", "which skills do I need", "update tool inventory". Produces docs/tool-routing-report.md and STOPS for human approval before any skills are moved.
How this skill is triggered — by the user, by Claude, or both
Slash command
/sentinel:routingThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Analyze the project context and the full global skill list, then produce a routing report that recommends which skills should be loaded into project scope. **This skill produces a document for human review — it does NOT move or copy any skills.**
Analyze the project context and the full global skill list, then produce a routing report that recommends which skills should be loaded into project scope. This skill produces a document for human review — it does NOT move or copy any skills.
Read the following files to understand what this project does:
CLAUDE.md (root)ARCHITECTURE.md or docs/architecture.md (if exists)README.md (if exists).claude/rules/tool-boundary.md (if exists — means boundary has run before)docs/tool-routing-report.md (if exists — means routing has run before; this is an UPDATE)Collect the full skill list from the current session's system reminder. For each skill, extract:
For each global skill, determine:
Provide a one-sentence rationale for each classification.
docs/tool-routing-report.mdIf the file already exists, update it (preserve developer notes from previous reviews, update the skill inventory and classifications). If it does not exist, create it.
Use this structure:
---
generated_at: {YYYY-MM-DD}
project: {project name from CLAUDE.md or directory name}
status: pending_review # pending_review | approved | needs_revision
total_skills_scanned: {count}
skills_included: {count}
skills_excluded: {count}
skills_uncertain: {count}
---
# Tool Routing Report
## Included Skills
| Skill | Source | Rationale |
|-------|--------|-----------|
| call-codex | user:skill | Project uses Codex for second opinions (see side_effect/codex_cli_usage.md) |
| ... | ... | ... |
## Excluded Skills
| Skill | Source | Rationale |
|-------|--------|-----------|
| original:latex-posters | plugin | Project is not academic poster work |
| ... | ... | ... |
## Uncertain — Needs Developer Decision
| Skill | Source | Rationale | Developer Decision |
|-------|--------|-----------|-------------------|
| original:pdf | plugin | Project has report docs but unclear if PDF generation needed | |
## Developer Notes
> {Space for developer to add notes during review. Preserved across updates.}
## Appendix: Full Global Skill List
{Complete list of all skills detected in this session, with source and description.}
Tell the developer:
docs/tool-routing-report.mdstatus: approved when ready/boundary to generate tool boundary declarationsDo NOT copy or move any skills. Do NOT modify .claude/skills/ or .claude/rules/. This skill only produces the report.
## Developer Notes sectionDeveloper Decision entries already filled instatus to pending_review and update generated_atnpx claudepluginhub touricks/fanshi_personal_skills --plugin sentinelAudits a project's recurring workflows from past sessions and memory, then recommends high-value skills to add or update.
Restructures oversized single-file Skills or scattered project rules into a well-organized Skill directory with progressive rigor tiers.
Generates .dh/skill_discovery.yaml by scanning repo for tech stack signals like pyproject.toml and package.json, inventorying skills via npx skills list, suggesting candidates, and writing config for skill injection.