More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    ChatGPT Chat Coding Agent — ccc-mcp

    1

    Turn ChatGPT Chat into a local coding agent with ccc-mcp. Edit code, run real tests, use MCP tools and Skills, and resume multi-agent workflows with verifiable evidence.

    Free

    Security scanned
    ChatGPT Chat Coding Agent — ccc-mcp

    ChatGPT Chat Coding Agent — ccc-mcp

    Example session with this skill installed

    Use ccc-mcp in ChatGPT-only mode to investigate and fix the SWE-bench Verified issue astropy__astropy-12907 in my authorized repository.

    Follow the SA → SD/RD → REVIEW workflow, make the necessary code changes, run pytest, and verify the final patch.

    Preserve the Git diff, execution history, test results, and exit codes. Do not report success unless supported by actual test evidence.

    • Read your context and instructions
    • Compiled the chatgpt chat coding

    SWE-bench Verified — astropy__astropy-12907

    Result: SOLVED

    This example summarizes a recorded ccc-mcp ChatGPT-only benchmark run. It demonstrates a structured coding workflow with actual repository changes, test execution, and independent evaluation evidence.

    Workflow

    The task followed the ccc-mcp development workflow:

    SA → SD/RD → REVIEW

    The workflow included bug investigation, code implementation, test verification, and review of the resulting changes.

    Verification Results

    VerificationResult
    Original baseline11 passed, 11 failed
    Tests after the fix22 passed, 0 failed
    Final pytest exit code0
    Git patch4 files, 4,312 bytes
    SWE-bench evaluationSOLVED
    Evaluation result1/1 resolved

    Execution Evidence

    The baseline test run reproduced the original failure.

    After the implementation was corrected, the targeted test run completed with all 22 tests passing and an exit code of 0.

    A Git patch was generated from the repository changes and evaluated using the SWE-bench harness. The evaluation reported that the issue was resolved.

    Evaluation run: ccc_arm_test_04

    What ccc-mcp Demonstrated

    • Coordinated development through the ChatGPT-only workflow.
    • Maintained task context and workflow progress.
    • Tracked repository changes through Git.
    • Retained actual test execution results and status information.
    • Supported evaluation with independently recorded evidence.

    Outcome

    The selected issue was successfully resolved with passing targeted tests and a successful SWE-bench harness result.

    This example illustrates ccc-mcp's ability to support structured AI-assisted software development with recoverable context and verifiable execution evidence.

    Project: https://github.com/bruce601080102/ccc-mcp

    Install: pip install -U ccc-mcp

    This is a formatted summary of a recorded benchmark case, not a live agent transcript or a claim about the entire SWE-bench Verified benchmark.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    ccc-mcp — AI Coding with Verified Tests

    Stop trusting "tests passed." Verify what your AI coding agent actually did.

    ccc-mcp is an open-source MCP server that connects ChatGPT to an authorized local coding workspace. It helps AI agents work through structured development tasks while preserving project context, actual code changes, test results, and review evidence.

    This Agensi Skill teaches your AI agent when and how to use ccc-mcp.

    Key Features

    • ChatGPT-first coding: Use ChatGPT to work on authorized local repositories.
    • Multi-agent orchestration: Optionally assign workflow roles to Codex CLI and Claude Code.
    • Structured workflow: Follow SA → SD/RD → REVIEW for analysis, implementation, testing, and review.
    • Persistent context: Save project state and resume unfinished tasks across conversations.
    • Verified test evidence: Record host-executed pytest results, exit codes, Git diffs, and execution history.
    • Local management UI: Manage projects, permissions, connections, tasks, and context.
    • Recoverable tasks: Inspect failures and continue from the recorded workflow state.

    Installation

    Requires Python 3.11+ and Git.

    Run these commands on your own computer:

    pip install -U ccc-mcp
    ccc-mcp ui
    

    Then:

    1. Open the local ccc-mcp management UI.
    2. Create or import a project.
    3. Authorize access to the selected repository.
    4. Configure the MCP connection for ChatGPT.
    5. Select the chatgpt-only profile or configure optional agent roles.
    6. Verify that the ccc-mcp tools are connected before starting a coding task.

    How This Agensi Skill Works

    Adding this Skill through Agensi provides instructions to your AI agent.

    It does not automatically install ccc-mcp, connect to your local computer, or grant filesystem permissions.

    The local ccc-mcp MCP server must be installed and connected separately.

    If the required tools are unavailable, the Skill guides the user through setup instead of claiming to have executed code or tests.

    Who Is It For?

    Developers who want to use ChatGPT and optional coding agents on real repositories while preserving context, tracking development progress, and verifying execution evidence.

    Documentation

    GitHub: https://github.com/bruce601080102/ccc-mcp

    Python package: https://pypi.org/project/ccc-mcp/

    License: MIT

    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 a separately installed and configured local ccc-mcp MCP server.
    • Connecting through Agensi does not automatically enable local code execution.
    • ChatGPT requires a supported MCP connection and an authorized project.
    • Optional Codex and Claude Code workflows require separately configured native CLI tools.
    • Test results depend on the target project's environment and dependencies.
    • Successful execution is only reported when actual tool and test evidence is available.

    How to use

    1. 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. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 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

    Free to use

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 15 days ago

    • Free with an Agensi account

    Accounts
    OpenAIAnthropic
    Software
    ccc-mcp MCPPython 3.11+GitOpenAI Platform Tunnel

    The OpenAI Platform Tunnel connects ChatGPT.

    Permissions

    Network Access
    Terminal / Shell
    Write Files
    Read Files

    Allowed Hosts

    api.openai.com
    api.anthropic.com

    This skill requires a separately installed and connected ccc-mcp MCP server. Terminal access is used for local installation, development commands, and test execution. File access is limited to repositories explicitly authorized by the user. Network access supports the separately configured MCP connection. The skill does not automatically grant filesystem access or retrieve runtime credentials. Credentials must be configured securely in the user's local environment.

    Listed16 days ago
    Updated14 days ago

    What's inside

    Frequently Asked Questions

    Yes. Install ccc-mcp on your own computer using pip install -U ccc-mcp, then run ccc-mcp ui to open the local management interface. Configure the MCP connection before starting development tasks.

    Yes. The chatgpt-only profile supports the development workflow without installing Codex or Claude Code. A separate ccc-mcp connection and project authorization are still required.

    ccc-mcp records workflow progress, repository changes, host-executed tests, and review evidence. It distinguishes actual execution results from an AI agent's unverified claims.

    New in Agents & Orchestration

    Free