learn

A process for turning a problem noticed during a chat into a lasting change to an agent's rules, skills, or automation hooks. It also requires fixing the immediate omission in the current conversation.

In plain words
What is it for?
It is for improving agent behaviour after an omission, choosing whether to update a rule, create or change a skill, or modify a hook.
Why use it?
It helps prevent the same behaviour from being missed again by recording the correction in the smallest suitable automation component.

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

Made for: Claude Code, Codex.

Per session 25 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,523 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 $0.00025 $0.01523
Opus 5 $0.00013 $0.00762
Sonnet 5 $0.00005 $0.00305
Haiku 4.5 $0.00003 $0.00152

Measured 2d ago against content hash 3f16e605a860, 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.

.claude/skills/learn/SKILL.md · 102 lines

How it starts

The opening of the file, as written. The whole thing — 102 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Learn

Turn a behavior miss raised in chat into a permanent process improvement.

Use this skill when the user points out an omission, asks why something was not done, or requests stronger automatic agentic behavior.

Goals

  1. Fix the immediate omission in the current turn.
  2. Prevent recurrence by updating the smallest durable artifact (rule, skill, or hook).
  3. Apply the new guidance immediately in the given chat context—perform the newly required behavior in this conversation, not only document it.

Inputs

  • issue_statement (required): What behavior was missed.
  • scope_hint (optional): repo, local_mcp, or specific path.
  • target_behavior (optional): Desired default behavior going forward.

Workflow

  1. Capture the failure clearly

    • Restate: expected behavior, actual behavior, and impact.
    • Confirm whether the missed action was not explicitly requested earlier in the turn flow (retrospective improvement case).
  2. Select the right remediation artifact

    • Rule update when behavior should be globally or repeatedly enforced.
    • Skill update/new skill when behavior is a multi-step workflow.
    • Hook/script update when behavior must be automatically enforced by tooling (for example, pre-commit checks or setup hooks).
    • MCP source-repo instruction update when omission is due to MCP instruction quality or ordering.
  3. Choose the repository to change

    • Current repo (ateles) for local agent workflow defaults, repo rules, and repo skills.
    • Relevant local MCP source repo when the issue is MCP-instruction behavior (for example Neotoma instruction text in sibling repo).
    • Priority rule: If the miss concerns Neotoma interactions, default to improving Neotoma's own instruction source first; use ateles only for non-MCP workflow behavior.
    • Prefer the smallest repository scope that fixes the issue for future runs.
    • Never add or modify ateles rules (or skills/hooks) to compensate for behavior that should be enforced by MCP server instructions. If the root cause is MCP instruction quality or ordering, fix the MCP source repo only; do not duplicate or compensate in this repo.

Read the full file on GitHub · 102 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 · 102 lines · 25 tokens per session scan A 3f16e605a860

Subscribe to this mod's changes

learn is a skill published in the GitHub repository markmhendrickson/ateles (5 stars, last pushed 5d ago), licensed MIT. It adds 25 tokens to every session and 1,523 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 2026-08-31.

Related

Other skills, from other repositories

share-skills

Use when you need to author, package, or share a skill folder — understanding the SKILL.md format, where skills live, and how to bundle a folder so it can be shared with others.

spinabot/brigade · 44 tokens

arbor-agent-executor

Executor-dispatch phase for Arbor. Use when implementing an Idea Tree node through RunExecutor or RunExecutorParallel semantics: isolated git worktree, executor prompt construction, eval metadata injection, RunTraining policy, smoke/full evaluation, report parsing, artifact persistence, tree update, and insight…

RUC-NLPIR/Arbor · 63 tokens

arbor-agent-plugins-hitl-budget

Domain adaptation, human-in-the-loop, and budget policy phase for Arbor. Use when a run mentions plugins, plugin profiles, mlekaggle, evalcontract, protectedpaths, requiredoutputs, lifecycle hooks, convergence, budgetpolicy, RunTraining stages, AskUser, or interaction modes auto/direction/review/collaborative.

RUC-NLPIR/Arbor · 76 tokens

bernstein-run

Run a verified multi-agent goal with Bernstein. Use when a task is too large for a single agent session: Bernstein decomposes the goal into tasks, spawns CLI coding agents in parallel git worktrees, verifies their output, and merges results. Also use to check run status, costs, and to verify a finished run against its…

sipyourdrink-ltd/bernstein · 75 tokens

dynamic-tools

Runtime tool management with toolmanage and tools.toml format. Create, enable, disable, reload tools without restart. (/dynamic-tools, toolmanage, runtime tools).

adolfousier/opencrabs · 36 tokens

bernstein-alerts

Show active alerts from Bernstein - failed tasks, stalled agents, budget warnings, blocked tasks needing human intervention. Use when the user asks about problems, errors, warnings, or what needs attention.

sipyourdrink-ltd/bernstein · 43 tokens