Help us improve
Share bugs, ideas, or general feedback.
From mapcoder
Solve coding problems using multi-agent retrieval, planning, coding, and debugging pipeline. Use when solving algorithmic problems, implementing features from specifications, or when code needs iterative refinement.
npx claudepluginhub newjerseystyle/plugin-map-coderHow this skill is triggered — by the user, by Claude, or both
Slash command
/mapcoder:mapcoderThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
You are orchestrating the MapCoder pipeline, a multi-agent system that replicates the human programming cycle through four specialized agents:
Coding workflow covering discovery, planning, implementation, and verification. Invoke whenever task involves any interaction with code — writing, modifying, debugging, refactoring, or understanding codebases. Runs discovery protocol before language-specific skills engage.
Delegates coding tasks (debug, implement, refactor) to OpenAI Codex CLI via codex exec, skipping the Node companion runtime for faster execution. Codex writes code; Claude verifies.
Leverages OpenAI Codex/GPT models for autonomous code implementation, reviews, and sandboxed task execution. Triggers on 'codex', 'use gpt', 'full-auto' etc.
Share bugs, ideas, or general feedback.
You are orchestrating the MapCoder pipeline, a multi-agent system that replicates the human programming cycle through four specialized agents:
Parse $ARGUMENTS to extract:
--lang <language>: Target programming language (default: Python)--sandbox: Use Docker sandbox for code execution (safer)Example inputs:
/mapcoder implement binary search → Python, direct execution/mapcoder --lang javascript implement binary search → JavaScript/mapcoder --sandbox --lang rust implement linked list → Rust, sandboxedFirst, analyze the input:
Use the Task tool to spawn the retrieval agent:
Spawn a retrieval agent to generate 3-5 similar problems for:
[problem description]
The agent should return:
- Similar problem descriptions
- Solution patterns/approaches used
- Key algorithmic concepts
Use subagent_type: "general-purpose" with the retrieval-agent.md system prompt.
Use the Task tool to spawn the planning agent with the retrieved examples:
Spawn a planning agent to create step-by-step plans for:
[problem description]
Using these similar problems as reference:
[retrieved examples]
Generate 2-3 alternative algorithmic plans.
Use the Task tool to spawn the coding agent:
Spawn a coding agent to implement the solution in [language]:
[problem description]
Following this plan:
[selected plan]
Test against these sample cases:
[test cases]
If --sandbox flag is set, instruct the agent to use scripts/sandbox-runner.sh instead of direct execution.
After coding completes:
If tests pass: Return the solution with explanation.
If tests fail: Enter debugging loop (max 3 iterations):
Spawn a debugging agent to fix the failing code:
Original problem: [problem]
Current code: [code]
Error output: [errors]
Original plan: [plan]
Identify the bug and generate corrected code.
When successful, output using the solution template:
## Solution
**Language**: [language]
**Status**: [Passed/Failed after N attempts]
### Code
[final code]
### Explanation
[step-by-step explanation of the approach]
### Test Results
[test output]
The pipeline supports adaptive routing: