learn

A learning-note tool that records what surprised the team, what happened, why it happened, and what to change next time.

In plain words
What is it for?
Use it after investigations, repeated fix attempts, new infrastructure work, or other tasks that produce lessons worth storing.
Why use it?
It turns notable development experiences—such as unexpected behaviour or difficult fixes—into reusable notes.

Skill for Claude CodeCodex

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/firatcand/forge/learn
Any agent
npx skills add firatcand/forge --skill learn
Clone the repo
git clone --depth 1 https://github.com/firatcand/forge

Made for: Claude Code, Codex.

Per session 28 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,046 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00028 $0.01046
Opus 5 $0.00014 $0.00523
Sonnet 5 $0.00006 $0.00209
Haiku 4.5 $0.00003 $0.00105

Measured 2d ago against content hash d4c515bae789, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

learn scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

skills/learn/SKILL.md · 112 lines

How it starts

The opening of the file, as written. The whole thing — 112 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/learn

Delegate to learning-curator.

Triggers (any one makes the task "notable")

  • Investigation took > 30 min
  • 2 fix attempts before success

  • Surprised by behaviour
  • Found a non-obvious gotcha
  • Made a non-trivial trade-off
  • Bootstrapped something new (test framework, CI, infrastructure)

Process

  1. Read the last commit + PR description + investigation file (if exists).

  2. Extract:

    • What we expected
    • What actually happened
    • Why
    • What we'd do differently
  3. Tag with relevant types.

  4. Resolve the canonical learnings path. docs/learnings/ is gitignored (forge-dogfood publish-hygiene rule), so its single source of truth is the main checkout's docs/learnings/ tree — not the working directory. /pickup-task hydrates worktrees by cp -r from the main checkout (skills/pickup-task/SKILL.md lines 47–53), and /learn must mirror that contract on the write side. See spec/SPEC.md §Learnings store for the canonical-store rule and why.

    Resolve the main checkout's absolute path via git rev-parse --git-common-dir (which always resolves to the main checkout's .git from anywhere — main or worktree). Compare against pwd -P so symlinked paths (e.g. macOS /var/private/var) don't trigger a spurious double-write:

    GIT_COMMON_DIR="$(git rev-parse --git-common-dir)"
    MAIN_ROOT="$(cd "$(dirname "${GIT_COMMON_DIR}")" && pwd -P)"
    PWD_REAL="$(pwd -P)"
    QUARTER="2026-Q2"   # or current quarter, e.g. "$(date -u +%Y)-Q$((($(date -u +%m)-1)/3+1))"
    SLUG="kebab-case-slug-from-title"
    mkdir -p "${MAIN_ROOT}/docs/learnings/${QUARTER}"
    
  5. Refuse on collision. If ${MAIN_ROOT}/docs/learnings/${QUARTER}/${SLUG}.md already exists, stop and surface the conflict — pick a different slug, or Edit the existing learning instead of writing a new one. Do not silently overwrite a prior learning.

  6. Write the canonical record first to the main checkout's absolute path using the Write tool. This is the load-bearing write; do not skip it even on errors elsewhere:

Read the full file on GitHub · 112 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 112 lines · 28 tokens per session scan A d4c515bae789

Subscribe to this mod's changes

learn is a skill published in the GitHub repository firatcand/forge (13 stars, last pushed 1mo ago), licensed MIT. It adds 28 tokens to every session and 1,046 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.