skills.rest
The-Vibe-Company
The-Vibe-CompanyOFFICIAL
Explore all skills by this author →
WebsiteGitHub

capture-learning-tools

Convert bugs, CI failures, and review feedback into project tests, CI checks, and agent instructions.

2.4k|292|Updated Feb 7, 2026
Search Tags:#regression testing#Documentation#ci#Agents.md#learning loop#GitHub actions#Code review
One-click install
npx skills add https://github.com/The-Vibe-Company/companion --skill capture-learning-tools
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: capture-learning-tools
Source: https://github.com/The-Vibe-Company/companion/tree/main/.agents/skills/capture-learning-tools
Command: npx skills add https://github.com/The-Vibe-Company/companion --skill capture-learning-tools

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

Lessons from bugs, CI failures, and code reviews often stay trapped in chat threads or one agent's private memory, so the same mistakes repeat. This Skill inspects the conversation and the repository, then promotes each durable lesson into the right prevention layer: shared agent instructions, regression tests, CI checks, scripts, docs, or ADRs.

Core Features & Use Cases

  • Learning Extraction: Analyzes conversations, bug reports, review comments, and CI failures to identify the failure mode and the earliest point it could have been caught.
  • Prevention Layer Selection: Uses a promotion matrix to decide whether a lesson belongs in AGENTS.md/CLAUDE.md, a regression test, CI wiring, a script, an ADR, or should be left manual.
  • Repository Inspection Scripts: Ships dependency-free Python scripts that audit AGENTS.md/CLAUDE.md symlinks and imports, summarize GitHub Actions workflows and CI cost signals, and classify lessons into likely destinations.
  • Use Case: After a content frontmatter bug slipped through review, run the learning loop to get a plain-English verdict, a proposed schema validation test, a budget-aware CI recommendation, and exact AGENTS.md rule text before any file is edited.

Quick Start

Review this conversation about the CI failure we just fixed and recommend what should change in the project so it does not happen again.

Frequently Asked Questions about capture-learning-tools

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

FAQPage Schema
How do I turn a bug fix into a permanent project improvement?▼

Run the learning loop on the conversation or incident. It identifies the failure mode, picks the earliest detection point, and recommends a regression test, CI check, script, or AGENTS.md rule, proposing changes before editing anything.

How to share agent instructions between Claude Code and Codex?▼

Create AGENTS.md as the shared source of truth and convert CLAUDE.md into a small adapter that imports it or symlinks to it. The inspect_project_guidance.py script detects this portability gap and checks for duplicated rules.

Should I add a CI check or a local script for a repeated failure?▼

It depends on repository cost mode. Public open-source repos can afford standard PR checks, while private repos should start with a local preflight command and fast path-filtered PR checks, keeping slow suites nightly or release-gated.

When should a lesson not be automated?▼

One-off judgment calls, subjective preferences, and unstable exploratory behavior should stay manual. The skill explicitly lists items not worth automating and explains why, rather than turning every mistake into a rule.

Does this skill edit my repository files automatically?▼

No. It proposes changes first with exact instruction text, test plans, and an apply plan. Files are only edited when you explicitly ask, and it never stages, commits, pushes, or changes billing-sensitive CI without approval.

Related Skills

Explore More Software Engineering Skills→
OmniNode-aiOmniNode-ai
3

pr-review-dev

Consolidate PR comments and CI failures, then dispatch agents to fix issues.

Official
Advanced
comeritocomerito

code-review

Identify changed files in a pull request and assess them against architecture, security, and quality rules.

Official
Advanced
AnveioAnveio
1

task-verification-pipeline

Run and debug repository verification pipelines with ./verify.sh and E2E modes.

Community
Intermediate
leocostaleocosta
4

enforce-precommit

Install and maintain pre-commit framework configs aligned with Octopus guardrails.

Community
Advanced