self-learning

Capture verified workflows from the current session as reusable Agent Skills.

Updated Aug 6, 2026
One-click install
npx skills add https://github.com/ferrarifankid04/ai-skills-public --skill self-learning-ferrarifankid04
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: self-learning
Source: https://github.com/ferrarifankid04/ai-skills-public/tree/main/claude-code/skills/self-learning
Command: npx skills add https://github.com/ferrarifankid04/ai-skills-public --skill self-learning-ferrarifankid04

SYSTEM DOCUMENTATION & REQUIREMENTS

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

What problem does it solve? Hard-won operational knowledge — how to reach the dev database, deploy, run migrations, or verify a change live — is rediscovered from scratch every new session. This Skill turns a proven "golden path" from the current conversation into a reusable Agent Skill so future sessions start already knowing it. ## Core Features & Use Cases - Automatic trigger recognition: Detects harvestable moments such as multi-attempt debugging, rediscovered project facts, recurring operational workflows, or explicit user requests like "remember this". - Promotion rule gating: Only writes a skill when there is a passing verification check, a named failure pattern, and at least one ruled-out dead-end, keeping confident guesses out of the skill set. - Guided authoring workflow: Distills the golden path, dedupes against existing skills, chooses project vs global scope, and delegates the write to a subagent or inline using a strict brief with a template and authoring spec. - Use Case: After spending an hour figuring out the exact command sequence to run migrations against the staging database, the Skill captures that procedure — including the dead-ends that failed — as a project-local skill committed to the repo for the whole team. ## Quick Start Ask the agent to save the workflow you just completed as a reusable skill, for example by saying "save this deployment procedure as a skill so I don't have to re-explain it next time".

Frequently Asked Questions about self-learning

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

FAQPage Schema
How do I save a workflow as a reusable agent skill?

Tell the agent to remember or save the workflow, or let it detect the moment automatically after non-trivial debugging. It distills the proven procedure, picks a project or global scope, and writes a new SKILL.md following the bundled template and authoring spec.

What is the promotion rule for creating a new skill?

A skill is only written when three conditions hold: a passing verification check such as a green test or clean command exit, a named failure pattern the path avoids, and at least one ruled-out dead-end approach. Otherwise the lesson is recorded as a tentative memory note or skipped.

Does this skill work with Claude Code and Codex?

Yes, it is tool-neutral and works with any agent that reads the Agent Skills format, including Claude Code and Codex, which both load SKILL.md skills natively. It delegates to a subagent where supported or works inline otherwise.

When should a lesson become a skill instead of a memory note?

Multi-step reusable procedures like deploys or database access become skills, while single standalone facts or one-line corrections belong in a lightweight memory index if one exists. Genuinely one-off tasks unlikely to recur should be skipped entirely.

Can secrets or credentials be stored in a generated skill?

No, secret values such as passwords, tokens, and connection strings must never be written into a skill file because skills get committed and shared. The skill records only where the secret lives, such as an env var name, selector function, or secret manager entry.