learn

learn is a command for Claude Code from terrene-foundation/metis. It costs 0 tokens per session (710 once invoked), scanned A, original, Apache-2.0.

A learning-status command for reviewing corrections, rule violations, completed work, decisions, and past codification in a coding project.

In plain words
What is it for?
Use it to view the learning digest, check observation statistics, and review what the codification process has already handled.
Why use it?
It brings useful lessons from earlier sessions into one summary, so recurring mistakes and decisions are easier to spot.

Command for Claude Code

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 commands/terrene-foundation/metis/learn
Clone the repo
git clone --depth 1 https://github.com/terrene-foundation/metis

Made for: Claude Code.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for learn

README.md
[![agentmods](https://agentmods.dev/badge/commands/terrene-foundation/metis/learn.svg)](https://agentmods.dev/commands/terrene-foundation/metis/learn)
Your own site
<a href="https://agentmods.dev/commands/terrene-foundation/metis/learn"><img src="https://agentmods.dev/badge/commands/terrene-foundation/metis/learn.svg" alt="Measured on agentmods" height="20"></a>
Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 710 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 $0.00000 $0.00710
Opus 5 $0.00000 $0.00355
Sonnet 5 $0.00000 $0.00142
Haiku 4.5 $0.00000 $0.00071

Measured 3d ago against content hash ba378d9ca780, 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 3d 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.

.claude/commands/learn.md · 68 lines

How it starts

The opening of the file, as written. The whole thing — 68 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/learn - Learning System Status

Purpose

View the learning digest and codification history. The learning system captures meaningful signals (user corrections, rule violations, session accomplishments, journal decisions) and feeds them into /codify for integration into real artifacts.

Quick Reference

Command Action
/learn Show learning digest summary
/learn stats Show observation statistics and breakdown

Usage

View Learning Digest

Read .claude/learning/learning-digest.json and present:

  1. Corrections — Times the user pushed back or redirected. These are the most valuable signals — each represents a gap in the current artifacts.
  2. Error patterns — Recurring rule violations (which rules are being violated most?).
  3. Accomplishments — What was completed in recent sessions.
  4. Decisions — Journal entries (DECISION, DISCOVERY, TRADE-OFF) that may need codification.
  5. Active frameworks — Which Kailash frameworks are in use.

View Codification History

Read .claude/learning/learning-codified.json to see what /codify has already processed from the digest.

View Stats

node scripts/learning/digest-builder.js --stats

How It Works

  1. Hooks capture signals — User corrections (UserPromptSubmit), rule violations (PostToolUse), session accomplishments (SessionEnd), journal decisions (SessionEnd). Pure file I/O, no LLM.
  2. Digest builder aggregates — At session end, observations are summarized into learning-digest.json. Pure aggregation, no pattern matching or confidence scores.
  3. /codify does the thinking — When /codify runs, the LLM reads the digest, journals, and session notes. It decides what to codify into real rules, skills, or agents. No intermediate staging — changes go directly into canonical artifact locations.

File Locations

<project>/.claude/learning/
  observations.jsonl        # Raw observations (capped at 500, auto-archived)
  observations.archive/     # Archived observations
  learning-digest.json      # Structured summary for /codify
  learning-codified.json    # What /codify has already processed

Read the full file on GitHub · 68 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. 3d ago First seen · 68 lines · 0 tokens per session scan A ba378d9ca780

Subscribe to this mod's changes

learn is a command published in the GitHub repository terrene-foundation/metis (2 stars, last pushed 4mo ago), licensed Apache-2.0. It costs nothing until one of its globs matches a file; then it loads 710 tokens. 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 2026-08-31.