Support the Project

Why community scenarios matter

MCP Steroid improves by learning from real-world usage. The more diverse the repositories, languages, and workflows we test against, the more reliable the product becomes for every user.

If your repository works in an IntelliJ-based IDE, it is in scope. Support the project and share your scenarios with us.

Usage logs sharing

Beyond full scenario submissions, you can help by sharing your tool call logs. The .idea/mcp-steroid folder in your project contains a log of every tool call your AI agent sent to the plugin. This data helps us fine-tune prompts, skills, and documentation to make agents more effective.

For details on how we process this data, see Learning Methodology.

Project scenarios sharing

Please provide:

  1. Repository pointer – public repo, temporary private access, or archive
  2. Commit SHA – exact revision to reproduce from
  3. Agent client/model/version – the exact agent setup used
  4. Baseline setup – toolchain and prompt flow without MCP Steroid, ideally a Docker container.
  5. Task prompt – the exact prompt given to the agent
  6. Review prompt – how the output was evaluated
  7. Acceptance criteria – binary checks and required artifacts
  8. Environment constraints – OS, IDE build, branch policy, time limits
  9. Data-handling constraints – public, anonymized, or private with restrictions

Submit via GitHub Issues.

How we process submissions

  1. Open the repository in an IntelliJ-based IDE
  2. Connect MCP Steroid to a supported agent
  3. Run baseline (without MCP Steroid) and treatment (with MCP Steroid) using orchestrated sub-agents
  4. Capture run artifacts and compare completion rate, manual interventions, regressions, token cost, and wall-clock time
  5. Convert stable findings into documentation, prompts, and benchmark tasks

Methodology details: Learning Methodology.

What you get back

  • Run artifact package with full logs
  • Baseline vs. treatment pass/fail report
  • Clear list of limitations encountered (if any)
  • Follow-up action items tailored to your workflow

How you can support

  • Developers: submit scenarios and issue reports with minimal reproductions
  • Engineering leaders: request pilot evaluations on your repositories – we are eager to learn alongside you
  • Sponsors and investors: support benchmark expansion and productization