- Home
- Skills
- Agents & Orchestration
- Lean Routing
Lean Routing
Help AI coding agents avoid unnecessary exploration, prioritize focused tools, and verify engineering tasks efficiently.
Free
Works with the AI Agents you already use
Lean Routing
Example session with this skill installed
I need to rename the 'user_id' field to 'account_id' across the entire 'services/billing' directory and ensure the tests still pass. How should we route this?
- Read your context and instructions
- Compiled the lean routing
Recommended Routing Plan
Task: Rename user_id to account_id within services/billing while preserving existing behavior.
Suggested route: Focused tool-based exploration → bounded implementation → targeted verification.
Step 1 — Locate the affected code
Use deterministic tools to identify references before modifying files.
rg -n "user_id" services/billing
Inspect only the relevant implementation, tests, and interfaces. Check whether the field is also used in API contracts, database schemas, or serialized data.
Step 2 — Choose the implementation path
If the change is local and its impact is understood, apply a minimal, scoped edit.
If a suitable delegated worker is available, bounded implementation may be assigned to it. Otherwise, continue directly with the current agent.
Do not perform an unrestricted repository-wide replacement.
Step 3 — Verify the change
- Review the diff for unintended modifications.
- Run the relevant billing tests.
- Check for remaining references.
- Escalate if the rename affects external contracts, database migrations, or backward compatibility.
Routing Decision
| Item | Decision |
|---|---|
| Initial exploration | Deterministic tools |
| Implementation | Scoped edit or available bounded worker |
| Verification | Targeted tests and diff review |
| Escalation | Required if public contracts or schema compatibility are affected |
Expected deliverable: A focused change, verification evidence, and any unresolved compatibility risks.
This is an illustrative routing plan, not a claim that the example code was executed.
Connects securely to your tools. The creator never sees your data.
About this skill
Lean Routing — Less Exploration. More Engineering.
Help AI coding agents choose the smallest reliable workflow for each engineering task.
Lean Routing provides practical rules for reducing unnecessary repository exploration, using deterministic tools first, and reserving deeper reasoning for problems that need it.
What you get
- Task routing: Choose between direct tools, bounded delegation when available, and deeper reasoning.
- Focused exploration: Use
rg, Git, and targeted file inspection instead of reading an entire repository. - Escalation rules: Recognize ambiguity, cross-system impact, and changes that require additional review.
- Mechanical verification: Check results with focused tests and deterministic tools.
- Cross-platform helpers: Optional Python, PowerShell, and Bash routing scripts.
- Reusable templates: Record routing decisions and verification evidence.
Practical use cases
Explore an unfamiliar codebase
Identify entry points, trace relevant dependencies, and avoid reading unrelated files.
Fix a small bug
Inspect the affected code, make a focused change, and run targeted tests.
Handle a complex change
Assess potential impact, identify unknowns, and decide whether deeper analysis or specialist assistance is needed.
What's included
- Agent-readable
SKILL.md - Routing and escalation policies
- Decision-manifest template
- Python, PowerShell, and Bash helper scripts
Compatibility
Designed for Claude Code, OpenAI Codex CLI, and Gemini CLI.
Python is not required for the core workflow. Helper scripts are optional, and execution depends on the tools available in your environment.
Important notes
Lean Routing provides engineering workflow guidance. It does not automatically switch AI models or guarantee token or time savings.
Free and open-source under the MIT License.
Routing scripts are keyword-based hints. Requires standard CLI tools like rg or git for deterministic paths. Fallback to parent occurs if named workers are not installed.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
1 use
Times used to date
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 1 month ago
- Free with an Agensi account