agentsclimarketplace

Skills router

Skill AravindS-Wick/aravindhan-skills/skills/skills-router

148+ curated Claude AI skills — agent orchestration, security, multi-repo PRs, design, mobile & more. One-command install.

Install
npx -y skills add AravindS-Wick/aravindhan-skills --skill skills-router

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Manage and execute tasks for skills-router. Use at the start of any development session to silently determine which skill applies — reads git context (staged files, changed files, branch) and message keywords, then auto-loads the right skill. Never asks the user which skill to use.

SKILL.md

3.6 KB, 904 tokens by cl100k_base, as published. Nobody here has run it

Skills Router

Automatically determines and loads the right skill based on project context. Zero user input required.

Decision Logic

digraph router {
  "Read context" [shape=box];
  "git diff staged?" [shape=diamond];
  "package.json changed?" [shape=diamond];
  "*.test.* changed?" [shape=diamond];
  "Sprint/planning?" [shape=diamond];
  "Pre-release?" [shape=diamond];

  "Read context" -> "git diff staged?";
  "git diff staged?" -> "load: global-security" [label="yes"];
  "git diff staged?" -> "package.json changed?" [label="no"];
  "package.json changed?" -> "load: global-tester --checks=npm-audit" [label="yes"];
  "package.json changed?" -> "*.test.* changed?" [label="no"];
  "*.test.* changed?" -> "load: global-tester --checks=jest,coverage" [label="yes"];
  "*.test.* changed?" -> "Sprint/planning?" [label="no"];
  "Sprint/planning?" -> "load: orchestrate --phase=sprint-start" [label="yes"];
  "Sprint/planning?" -> "Pre-release?" [label="no"];
  "Pre-release?" -> "load: orchestrate --phase=pre-release" [label="yes"];
  "Pre-release?" -> "load: dev-assistant (default)" [label="no"];
}

Context Signals → Skill Mapping

SignalAuto-load
Staged files in gitglobal-security --checks=git,linting
src/ files changeddev-assistant --feature=review
*.test.* changedglobal-tester --checks=jest,coverage
package.json changedglobal-tester --checks=npm-audit
*.scss/*.css changedproduct-designer --feature=consistency
PR URL in contextglobal-security --full
"sprint" / "plan" in messageorchestrate --phase=sprint-start
"launch" / "release" in messageorchestrate --phase=pre-release
"competitor" / "market"market-researcher
"metric" / "kpi" / "roi"biz-analyst
"user story" / "backlog"po-guide
"component" / "design"product-designer
New file with no test fileqa-automation --feature=unit-stubs
No signal → defaultdev-assistant

How to Run This

On session start, silently:

# 1. Read context
STAGED=$(git diff --name-only --cached 2>/dev/null)
CHANGED=$(git diff --name-only 2>/dev/null)
BRANCH=$(git branch --show-current 2>/dev/null)

# 2. Determine skills needed (see mapping above)
# 3. Load them via Skill tool
# 4. Announce only if skill changes behavior

Rules

  • Never ask "which skill do you want?" — determine it from context
  • Load silently unless the skill changes how you'll respond
  • One skill at a time unless phase requires parallel (orchestrate handles that)
  • Verify relevance before executing — loading ≠ acting
  • If context is ambiguous → load dev-assistant as safe default
  • If user explicitly names a skill → use that, skip routing

Verification Before Execution

After loading a skill, verify it applies:

  1. Does the skill's "Use when..." match current context? → Proceed
  2. Does it NOT match? → Unload, try next signal
  3. Still uncertain? → Load dev-assistant, state what you detected

Red Flags (stop routing, ask user)

  • Multiple conflicting high-priority signals
  • User's message explicitly contradicts detected context
  • Session is continuation of specific ongoing task

Token cost: 50-100 tokens (context read only, no analysis)

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.