- Home
- Skills
- Agents & Orchestration
- agent router
Works with the AI Agents you already use
agent router
A multi-agent coordination framework for planning, executing, and independently validating project deliverables.
$2
agent router
Example session with this skill installed
I need to refactor the payment gateway interface. Check the current status and tell me how we should proceed with the implementation and validation.
- Read your context and instructions
- Compiled the agent router
I have performed a read-only audit of the current repository state and the payment-gateway module.
Project Audit & Routing Plan
| Category | Finding |
|---|---|
| Current Status | Branch feature/payment-refactor has 3 uncommitted changes in types.ts. |
| Last Evidence | No existing validation records found for the new interface. |
| Risk Score | 5/8 (Mode B): High coupling with order-service and data integrity risk. |
| Role Assignment | Planner: GPT-4o (Reasoning) |
Proposed Execution Pipeline
- Planning: Freeze interface definitions and idempotency requirements.
- Construction: Implement the
IPaymentProviderinterface insrc/gateways/. - Independent Review: The Reviewer instance will run the integration test suite and verify mock edge cases.
- Handoff: Deliver versioned file hashes and test logs.
Next steps
- Confirm if I should proceed with Mode B execution.
- Provide authorization for the Builder to modify
src/gateways/. - Specify if any specific legacy payment providers should be excluded from this batch.
Connects securely to your tools. The creator never sees your data.
About this skill
Your AI agents often struggle with boundaries, either over-committing to unmapped tasks or failing to provide verified deliverables. This skill implements a structured coordination framework for Codex environments, ensuring every code change or asset delivery follows a strict protocol of planning, execution, and independent validation.
What it does
- Status Auditing Performs read-only checks of project state, Git status, and task logs without triggering side effects.
- Dynamic Role Mapping Discovers available models and maps them to Coordinator, Planner, Builder, or Reviewer roles based on real-time environment capabilities.
- Risk-Based Scoring Categorizes tasks (Lightweight, Standard, or High-Complexity) using a 0-8 point scale across requirements, coupling, risk, and failure history.
- Independent Validation Enforces a strict separation of concerns where the Builder and Reviewer must be different agent instances to prevent self-grading.
- Automated Repair Cycles Manages internal feedback loops where the Reviewer provides specific defect evidence and the Builder refines the implementation within the existing authorization scope.
How it works
- Discovery The agent scans for
AGENTS.mdand project docs to identify current state, active tasks, and available model roles. - Scoring You provide the goal; the skill scores the task and selects the appropriate execution mode (A or B) and model tiers.
- Execution Loop The Planner freezes criteria, the Builder implements and self-checks, and the independent Reviewer runs verification.
- Reporting You receive a concise summary including version fingerprints, evidence of testing, and the specific models used for each phase.
Frameworks & tools
Works with any project structure (Git or non-Git) within a Codex-compatible environment. It interfaces with references/roles.md for model management and uses local task logs or docs/status.md for persistence.
Why this beats prompting it yourself
Generic prompts often lead to agents "grading their own homework" or blindly repeating failed logic. This skill forces a formal hand-off between specialized roles and requires objective evidence before marking a task as complete, preventing silent failures.
Use cases
- Executing complex refactors that require cross-module verification.
- Managing multi-agent workflows where planning and execution must be decoupled for safety.
- Providing high-assurance project status reports and progress recovery after a session break.
- Troubleshooting recurring defects by triggering specialized diagnostic roles before re-attempting a fix.
Known limitations
Does not perform silent model swaps; if a required model is unavailable via the host tool, the process pauses. Read-only checks cannot verify remote Git state without explicit query tools.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Agents & Orchestration.
Requires references/roles.md to exist for role mapping. Builder and Reviewer must be separate instances. Read-only mode cannot trigger external writes or deployments.
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 buying a skill.
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
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 9 days ago
- Software
Git
Permissions
No special permissions declared or detected