skills.rest
openclaw
openclawOFFICIAL
Explore all skills by this author →
Website𝕏@openclawGitHub

convex-improve-convex-plugin

Sends coding session transcripts to the Convex review endpoint for AI post-mortem analysis.

9.4k|1.5k|Updated Jan 3, 2026
Search Tags:#transcript#post-mortem#convex#Quickstart#Feedback#consent#session review
One-click install
npx skills add https://github.com/openclaw/clawhub --skill convex-improve-convex-plugin
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: convex-improve-convex-plugin
Source: https://github.com/openclaw/clawhub/tree/main/.agents/skills/convex-improve-convex-plugin
Command: npx skills add https://github.com/openclaw/clawhub --skill convex-improve-convex-plugin

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Convex quickstart maintainers lack visibility into where AI coding agents get stuck when following the quickstart runbook, bootstrap script, and skills. This Skill closes that feedback loop by submitting the current session transcript for a structured AI post-mortem.

Core Features & Use Cases

  • Transcript Submission: Runs an anteater-served helper script that locates the Claude/Codex .jsonl transcript and POSTs it to the /review endpoint.
  • Consent-Gated Sharing: Sharing is opt-in; the helper prints CONSENT_REQUIRED until the user explicitly chooses Always, Just this once, or Never.
  • Structured Findings: Returns findings (ambiguous instructions, agent-stuck patterns, tooling failures, wins) with title, target, and suggestedFix, plus automatic redaction of keys and tokens before upload.
  • Use Case: After finishing a Convex quickstart session where the agent struggled with setup steps, submit the transcript so the Convex team receives a post-mortem identifying which runbook instructions confused the agent.

Quick Start

Ask the agent to send this session's transcript to the Convex team for review, then choose a sharing consent option when prompted.

Frequently Asked Questions about convex-improve-convex-plugin

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I send my coding session transcript to the Convex team?▼

Run the anteater-served helper with curl piping to bash, passing a one-line app idea. The script locates the Claude or Codex .jsonl transcript and submits it to the /review endpoint after you grant consent.

Is my consent required before the transcript is uploaded?▼

Yes, sharing is strictly opt-in. The helper prints CONSENT_REQUIRED and exits with code 4 until you explicitly choose Always, Just this once, or Never, and nothing is sent before that choice.

Are API keys and secrets redacted before transcript upload?▼

Yes, the helper script redacts keys and tokens before uploading the transcript. The returned summary also stays focused on system findings rather than user data.

What does REVIEW_NO_TRANSCRIPT mean when running the helper?▼

REVIEW_NO_TRANSCRIPT means no Claude or Codex .jsonl transcript file was found for the current session. The skill instructs the agent to inform the user rather than retry or fabricate a submission.

What kind of findings does the Convex transcript review return?▼

The review returns structured findings such as ambiguous instructions, agent-stuck patterns, tooling failures, and wins. Each finding includes a title, target (runbook, bootstrap script, skills, or components), and a suggestedFix.

Related Skills

Explore More Software Engineering Skills→
tz5514tz5514

research-transcript

Audit agent session transcripts for compliance and reasoning gaps.

Community
Advanced
AbsolutelySkilledAbsolutelySkilled
207

incident-management

Coordinate incident response workflows from detection to post-mortem.

Official
Advanced
nielsmadannielsmadan
1

review-logs

Analyze Claude Code session transcripts to identify failure patterns and suggest fixes.

Community
Advanced
openclawopenclaw
388.4k

agent-transcript

Generate sanitized agent transcript sections for GitHub PR and issue bodies.

Official
Intermediate