OpenSkillIndex← back to search
claude-code

Agent-skills-standard common-learning-log

Append a structured learning entry to AGENTS_LEARNING.md whenever an AI agent makes a mistake. Auto-activates as a composite skill when: a pre-write skill violation is detected and auto-fixed, or when the session retrospective finds a correction loop. Also triggers directly when the user corrects the AI mid-session. Use when: mistake, wrong, redo, that's not right, correction, my bad, fix that error, I made a mistake, agent error, learning log, log mistake, AGENTS_LEARNING.md (triggers: AGENTS_LEARNING.md, mistake, wrong, redo, correction, agent error, learning log)

install
source · Clone the upstream repo
git clone https://github.com/HoangNguyen0403/agent-skills-standard
Claude Code · Install into ~/.claude/skills/
T=$(mktemp -d) && git clone --depth=1 https://github.com/HoangNguyen0403/agent-skills-standard "$T" && mkdir -p ~/.claude/skills && cp -r "$T/.agent/skills/common/common-learning-log" ~/.claude/skills/hoangnguyen0403-agent-skills-standard-common-learning-log && rm -rf "$T"
manifest: .agent/skills/common/common-learning-log/SKILL.md
source content

Agent Learning Log

Priority: P1 (OPERATIONAL)

Write a structured mistake entry to 

AGENTS_LEARNING.md
in the project root before retrying any corrected action.

Protocol

  1. Detect signal — identify which surface triggered this skill: 
    • Pre-write violation
      — a 
      common-feedback-reporter
      violation block was emitted with 
      Auto-fixed: YES
    • User correction
      — user used correction language mid-session
    • Session retrospective
      — a correction loop was found during 
      common-session-retrospective
  2. Read 
    AGENTS_LEARNING.md
     — count existing 
    ## Agent Learning Log: Iteration
    headers → N
  3. Append entry — write Iteration #(N+1) using the format in Log Entry Format
  4. Continue — proceed with the corrected action (non-blocking)

Guidelines

  • One entry per correction event — not one per file or per task
  • Concrete mistakes only — name the specific file, rule, or action that was wrong
  • The "Better Approach" must be actionable — state what to do, not what to avoid
  • Create file if missing — bootstrap with the header from Log Entry Format
  • Never skip for "minor" corrections — all corrections are learning signals

Anti-Patterns

  • No vague mistakes: 
    "I made a mistake"
    → name the specific pattern or rule violated
  • No skipping the log: Even if already in a hurry to fix, append the entry first (it takes <10 seconds)
  • No duplicate entries: One correction event = one entry, even if multiple files were affected
  • No overwriting: Always append to the bottom; never edit past entries

References