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.
npx agentmods add skills/trtmn/agent-plugins/learningsnpx skills add trtmn/agent-plugins --skill learningsgit clone --depth 1 https://github.com/trtmn/agent-pluginsWhat 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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00082 | $0.01841 |
| Opus 5 | $0.00041 | $0.00920 |
| Sonnet 5 | $0.00016 | $0.00368 |
| Haiku 4.5 | $0.00008 | $0.00184 |
Grade A, and why
learnings 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 yesterday.
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.
How it starts
The opening of the file, as written. The whole thing — 139 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Learnings
Autonomous, in-session capture of everything worth remembering across sessions. This skill documents the contract; the actual work is done by the learnings subagent, invoked in the background by main Claude whenever a trigger fires.
This skill is the capture half of the learning loop. The review + auto-promote half lives in the self-improvement skill (same plugin). Capture is autonomous and continuous; review is autonomous on session-end (and runnable on demand).
Relationship to self-improvement
These are two skills, one plugin, one invocation model each — a clean seam:
learnings |
self-improvement |
|
|---|---|---|
| Who invokes | Main Claude, autonomously, on every trigger | The SessionEnd hook (headless) — or you, via /self-improvement |
| When | Continuously, as things happen | After a session ends (if enough piled up), or on demand |
| Does | Appends pending entries to ~/.learnings/ |
Sweeps, investigates "needed-ness", auto-promotes qualifiers to CLAUDE.md, logs to CHANGELOG |
| Behavior | Pure capture, never promotes | Non-interactive auto-promote, with a conservative bar + revertible audit trail |
Capture is cheap; promotion is gated. The two never share behavior, so neither gets invoked in the other's mode.
The Trigger Contract
Main Claude (not this skill directly) watches for these signals and delegates to the learnings subagent via the Agent tool, in the background (run_in_background: true):
- User corrections: "Actually, it should be...", "No, don't...", "That's wrong because..." → LEARNINGS.md
- Knowledge gaps: outdated info given, project convention missed, rule had to be taught → LEARNINGS.md
- Suggestions: user proposed a better approach, tool, or pattern → LEARNINGS.md
- Command/tool failures: shell command failed, API errored, file op behaved unexpectedly → ERRORS.md
- Feature requests: user asked for something Claude couldn't do → FEATURE_REQUESTS.md
- Repeated mistakes: same slip twice in one session → LEARNINGS.md, priority high
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.
- yesterday First seen · 139 lines · 82 tokens per session scan A 37f9e14ac7d2
learnings is a skill published in the GitHub repository trtmn/agent-plugins (2 stars, last pushed 5d ago), licensed Unlicense. It adds 82 tokens to every session and 1,841 once invoked, about $0.0004 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.
Other skills, from other repositories
skill-template
Template for creating new Agent Skills for context engineering. Use this template when adding new skills to the collection.
create-skill
Guides users through creating effective Agent Skills for Cursor. Use when the user wants to create, write, or author a new skill, or asks about skill structure, best practices, or SKILL.md format.
creating-skills
Guide for creating Claude Code skills following Anthropic's official best practices. Use when user wants to create a new skill, build a skill, write SKILL.md, update an existing skill, or needs skill creation guidelines. Provides structure, frontmatter fields, naming conventions, and new features like dynamic context…
aish-add-skills
Create new skills from documents, tutorials, or examples. Use when user wants to create a skill from learning materials or existing content.
master-debate
Use when user explicitly asks for an adversarial / multi-round dialectic between masters — 祖师辩论, 各执一词, 谁更对, debate, 应成 vs 顿悟, 顿渐之争. Differs from /compare-masters (parallel single-round) by being adversarial multi-round via fresh-subagent orchestration. Topics 空有 / 禅净 / 性相 / 戒律 vs 内观 — trigger is adversarial framing…
skill-master
Guides skill creation and updates with specialized knowledge and workflows. Use when: "создай скилл", "измени скилл", "гайд по скиллам", "обнови скилл", "улучши скилл", "create skill", "update skill", "skill guide", "new skill", "how to write a skill".