teach

A command for teaching the mathlib-quality skill a project-specific coding rule, preferred technique, or example. It stores guidance such as which tactic to try first or how files should be organized.

In plain words
What is it for?
Use it to record naming rules, preferred proof tactics, code structure, or approaches to avoid. It accepts a short instruction or can guide an interactive before-and-after example.
Why use it?
It prevents developers from repeating the same instructions during future work. The skill can retain conventions and successful problem-solving patterns for that project.

Command

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/cbirkbeck/mathlib-quality/teach
Clone the repo
git clone --depth 1 https://github.com/CBirkbeck/mathlib-quality
Per session 11 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 2,024 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 $0.00011 $0.02024
Opus 5 $0.00005 $0.01012
Sonnet 5 $0.00002 $0.00405
Haiku 4.5 $0.00001 $0.00202

Measured 2d ago against content hash 8ebc864ce1b9, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

teach 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.

commands/teach.md · 276 lines

How it starts

The opening of the file, as written. The whole thing — 276 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/teach - Teach a Pattern or Convention

Explicitly teach the mathlib-quality skill something it should remember for this project.

Usage

/teach "always use grind before omega for Fin goals in this project"
/teach "in this codebase, prefer explicit universe variables"
/teach  (interactive mode - will prompt for input)

What You Can Teach

Project-Specific Conventions

/teach "always use grind before omega for Fin goals"
/teach "in this project, helper lemmas go in the Aux namespace"
/teach "we use Module.lean not Defs.lean for definition files"

Golf Patterns Discovered

/teach "fun_prop closes Continuous goals about our custom FooMap after simp [FooMap]"
/teach "grind [bar_def] handles all Bar.card goals in one step"

Before/After Examples

/teach

Then provide a before/after code example when prompted.

Things to Avoid

/teach "don't use omega for goals involving Fin.castSucc - it loops"
/teach "aesop times out on goals with more than 3 Set.mem hypotheses in this project"

Workflow

Step 1: Parse the Teaching

Analyze the user's input to extract:

  1. The pattern or rule being taught
  2. The category (golf, style, naming, structure, avoidance)
  3. Any code examples (before/after if provided)
  4. The math area if identifiable

Step 2: Categorize

Determine the learning type:

Input Pattern Type Example
"always/prefer/use X" user_teaching "always use grind for Fin goals"
"don't/avoid/never X" user_teaching (negative) "don't use omega for Fin.castSucc"
Before/after code golf_pattern or style_correction Code example pair
"in this project..." user_teaching Project convention
Naming preference naming_fix "use _aux not _helper"

Step 3: Write the Learning Entry

Write a structured entry to .mathlib-quality/learnings.jsonl (create the file and directory if they don't exist):

{
  "id": "<generate a short unique id>",
  "timestamp": "<current ISO timestamp>",
  "command": "teach",
  "type": "user_teaching",
  "before_code": "<before code if provided, otherwise empty>",
  "after_code": "<after code if provided, otherwise empty>",
  "pattern_tags": ["<extracted pattern tags>"],
  "description": "<the teaching, preserved as close to the user's words as possible>",
  "math_area": "<detected area or 'other'>",
  "accepted": true,
  "source": "user_teaching",
  "context": {
    "file_path": "",
    "theorem_name": "",
    "project": "<project name from git remote if available>"
  }
}

Read the full file on GitHub · 276 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 · 276 lines · 11 tokens per session scan A 8ebc864ce1b9

Subscribe to this mod's changes

teach is a command published in the GitHub repository CBirkbeck/mathlib-quality (32 stars, last pushed 13d ago), licensed MIT. It adds 11 tokens to every session and 2,024 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.