dynamic-routing
Skill4 starsChanged 5 months agoApache-2.0View on GitHub
Autonomous intent detection, dynamic agent spawning, capability-flag auto-resolution, and contextual skill injection based on task requirements.
In this repository48 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- api-designSkill
- improve-codebase-architectureSkill
- architectureSkill
- cicdSkill
- cloud-infraSkill
- codebase-navigatorSkill
- code-qualitySkill
- code-reviewSkill
- code-standardsSkill
- configuration-managementSkill
- contract-testingSkill
- dependency-managementSkill
- diagnoseSkill
- documentationSkill
- enterprise-architectureSkill
- enterprise-test-suitesSkill
- git-guardrails-claude-codeSkill
- grill-meSkill
- handoffSkill
- intent-clarificationSkill
- triageSkill
- legacy-modernizationSkill
- llm-evalsSkill
- observabilitySkill
- ops-sreSkill
- performance-optimizationSkill
- to-issuesSkill
- playwrightSkill
- to-prdSkill
- setup-pre-commitSkill
- prototypeSkill
- pr-reviewSkill
- ralph-loopSkill
- requirementsSkill
- root-cause-analysisSkill
- security-auditSkill
- security-patchingSkill
- startup-lifecycleSkill
- tddSkill
- testingSkill
- threat-modelingSkill
- to-issuesSkill
- token-efficiencySkill
- to-prdSkill
- ux-designSkill
- write-a-skillSkill
- zoom-outSkill
The file
We split this skill into the rules it gives an agent
5 rules. About 0 are common, 1 is distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 15
When a user submits any natural language request, the system detects the primary intent and automatically routes execution:
- Very specific: line 36
If an agent encounters an untested API boundary → Automatically invoke contract-testing or enterprise-test-suites.
- Very specific: line 37
If a build step introduces a new third-party dependency → Automatically trigger security-patching to scan for CVEs.
- Very specific: line 38
If an error occurs during vertical slice implementation → Automatically engage root-cause-analysis and ralph-loop.
- Very specific: line 39
If context window exceeds token thresholds → Automatically invoke token-efficiency and trigger /handoff compaction.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Autonomous Intent Routing & Dynamic Agent Invocation Skill
This skill allows the system to autonomously analyze user requests, map them to the exact agents, commands, and skills needed, and dynamically invoke specialized workflows without requiring the user to memorize command syntax.
1. Autonomous Request Routing Matrix
When a user submits any natural language request, the system detects the primary intent and automatically routes execution:
| User Trigger Phrasing / Goal | Auto-Inferred Phase | Primary Agent | Automatically Injected Skills & Commands |
| "I have a new idea...", "Build a SaaS...", "Let's make an app..." | Plan & Founder | product-manager | intent-clarification, grill-me, requirements, startup-lifecycle, /sdlc-plan, /sdlc-startup |
| "Implement feature...", "Code this slice...", "Add database model..." | Build | fullstack-engineer | enterprise-architecture, token-efficiency, ralph-loop, contract-testing, llm-evals, /sdlc-build |
| "How does this codebase work?", "Find symbol in 1.5M files...", "Explain architecture..." | Navigate | fullstack-engineer | codebase-navigator, zoom-out, code-review-graph, /sdlc-navigate |
| "Debug this 500 error", "Why did calculation fail?", "Fix bug in policy engine" | Diagnose & RCA | qa-engineer / fullstack-engineer | root-cause-analysis, diagnose, token-efficiency, /sdlc-diagnose |
| "Audit security", "Patch CVEs", "Harden headers", "Implement Zero-Trust" | Security | security-engineer | security-patching, security-audit, threat-modeling, /sdlc-security |
| "Migrate legacy code", "Decouple monolith", "Extract stored procedure" | Modernize | fullstack-engineer / devops-engineer | legacy-modernization, architecture-refactor, /sdlc-modernize |
| "Run tests", "Write E2E test", "Load test this API", "Chaos test DB" | Test Matrix | qa-engineer | enterprise-test-suites, playwright, tdd, /sdlc-test |
| "Deploy to AWS/K8s", "Setup CI/CD", "Create Dockerfile", "Terraform infra" | Ship | devops-engineer | cicd, cloud-infra, precommit-hooks, /sdlc-ship |
| "Set up SLOs", "Configure alerts", "Track MRR & Churn", "Write runbook" | Operate | sre-engineer | ops-sre, observability, startup-lifecycle, /sdlc-operate |
2. Autonomous Invocation & Chaining Protocol
Agents should autonomously trigger downstream skills and commands when the context requires it:
- Context-Aware Triggering:
- If an agent encounters an untested API boundary → Automatically invoke
contract-testingorenterprise-test-suites. - If a build step introduces a new third-party dependency → Automatically trigger
security-patchingto scan for CVEs. - If an error occurs during vertical slice implementation → Automatically engage
root-cause-analysisandralph-loop. - If context window exceeds token thresholds → Automatically invoke
token-efficiencyand trigger/handoffcompaction.
- If an agent encounters an untested API boundary → Automatically invoke
- Zero-Friction User Experience:
- The user can type free-form natural language (e.g. "Fix the billing webhook bug and deploy a hotfix"), and the orchestrator will chain
/sdlc-diagnose→/sdlc-implement→/sdlc-test→/sdlc-shipautonomously with approval checkpoints.
- The user can type free-form natural language (e.g. "Fix the billing webhook bug and deploy a hotfix"), and the orchestrator will chain
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.