mcpField
C
convex-improve-convex-plugin

convex-improve-convex-plugin

github.com/get-convexAgent workflows6931.1k installsUpdated 1d ago
Repository

About convex-improve-convex-plugin

improve-convex-plugin is a skill for improving the Convex development system by submitting a coding-session transcript to anteater for an AI post-mortem, only with explicit consent. It returns structured findings about instructions, tooling issues, and wins, focusing on system materials rather than end-user data.

Key features

  1. 01Submit transcripts only with consent
  2. 02Receive structured post-mortem findings
  3. 03Identify ambiguous instructions and tooling failures
  4. 04Summarize key findings and wins
  5. 05Redact keys and tokens before upload

Use cases

  1. 01Find recurring tooling failures in coding sessions
  2. 02Identify improvements to runbooks or skill instructions
  3. 03Review findings to plan system improvements

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/get-convex/agent-skills --skill convex-improve-convex-plugin

After installing, type/convex-improve-convex-pluginin Claude Code to confirm it loaded.

Details

From the project README

improve-convex-plugin

Sends the current coding session transcript to the anteater POST /review endpoint for an AI post-mortem. The review returns structured findings (ambiguous instructions, agent-stuck patterns, tooling failures, wins) targeted at the runbook, bootstrap script, skills, and components — not end-user data. Sharing is opt-in: the anteater-served helper asks once (Always / Just this once / Never) and remembers the choice.

Workflow

  1. Run the anteater-served helper: curl -fsSL "<anteater>/send-transcript" | bash -s -- --idea "<one-line app idea from this session>".
  2. If it prints CONSENT_REQUIRED (exit 4), the user has not chosen yet — ask them to share Always, Just this once, or Never, then re-run appending --consent always|once|never. Do not send until they answer.
  3. Watch for output markers: REVIEW_SOURCE (transcript found), REVIEW_SUBMITTED id=... (accepted), REVIEW_DONE status=done (findings ready).
  4. Summarize the highest-severity findings for the user: title → target → suggestedFix, then wins. Keep the summary about the system, not the user's data.

Rules

  • Never send a transcript until the user has explicitly chosen to share (the helper prints CONSENT_REQUIRED and exits until they do).
  • REVIEW_NO_TRANSCRIPT means no Claude/Codex .jsonl was found — tell the user.
  • Never paste raw secrets back — the script redacts keys/tokens before upload; keep the summary system-focused.
  • This is a system-improvement loop, not end-user feature feedback.

Comments

Markdown supported