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 skills add boykush/scraps --skill learngit clone --depth 1 https://github.com/boykush/scrapsWrote 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.
[](https://agentmods.dev/skills/boykush/scraps/learn)<a href="https://agentmods.dev/skills/boykush/scraps/learn"><img src="https://agentmods.dev/badge/skills/boykush/scraps/learn/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/boykush/scraps/learn"><img src="https://agentmods.dev/badge/skills/boykush/scraps/learn.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00128 | $0.02782 |
| Opus 5.5 | $0.00051 | $0.01113 |
| Sonnet 5.5 | $0.00026 | $0.00556 |
| Haiku 4.5 | $0.00013 | $0.00278 |
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.
How it starts
The opening of the file, as written. The whole thing — 137 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Learn
Help the user understand a source, then record only what they understood as one page in the wiki.
The wiki is a record of the user's understanding, not of what its sources say. When an AI drafts a summary and the user approves it, nobody can tell what was understood from what was merely read and nodded at. So before anything is written, the user answers for themselves. The goal is the user's understanding, and the page is what remains of it. A run that ends with nothing written, because understanding was not reached, has ended correctly.
Hold the dialogue and write the page in the user's language.
Boundaries
- Write no file until the comprehension check is done.
- Write only what passed the check. Something you explained is not established until the user has shown they understand it.
- One run produces one page. If the source holds several concepts, pick the central one and name the rest as candidates for later runs.
- State as fact only what you verified in a primary source. Analogies and examples are fine as teaching aids, but build the questions and the page from what the source says.
- Do not write the user's answers for them. If you quietly fix a wrong answer and let it pass, the page records your understanding, not theirs.
- Change existing pages only by turning literal mentions the user approved into links.
- Do not widen the work into tidying or auditing the wiki.
Asking the user
Several steps put choices to the user. How depends on the client, so decide it from the tools you actually have:
- When the client has a tool for asking multiple-choice questions (for example
AskUserQuestionin Claude Code), use it. Its definition says how many questions a call takes, how many options a question takes, and whether several options can be picked; stay inside those. - When there is no such tool, or what you need to ask does not fit it, ask in the message body: number the questions, letter the options, say how to reply (for example
1b 2d), then stop and wait.
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.
- 2d ago Changed d933c3bc0e65
- 3d ago Changed · +12 lines 220734e4df38
- 4d ago Changed · +9 lines f23720603344
- 5d ago First seen · 116 lines · 128 tokens per session scan A 6f95f7d89d74
learn is a skill published in the GitHub repository boykush/scraps (45 stars, last pushed today), licensed MIT. It adds 128 tokens to every session and 2,782 once invoked, about $0.0005 per session on Opus 5.5. 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 2026-10-11.
Other skills, from other repositories
llmwiki
Build and use a persistent, citation-traceable knowledge wiki from documents, research papers, notes, or project documentation. Use for llmwiki projects, retrieving reusable evidence for an agent task, or compiling a corpus for repeated questions. A one-off file read or live web search does not need a wiki.
processing-markdown
Processes Markdown files using mq, a jq-like query language for Markdown. Use when the user mentions Markdown processing, content extraction, document transformation, or mq queries.
web-scraping
Fetches web pages and extracts structured data using mq's toolchain (mq-crawl for fetching/crawling/JS rendering, mq for HTML-to-Markdown selector-based extraction, http() for in-query requests). Use when the user wants to scrape a URL, pull structured data out of a webpage, crawl a site, or turn HTML into…
improve
Critique an outside idea (a pasted tip, post, transcript or video summary) against what this system already has. Use when the founder pastes something and asks whether to adopt it, or says 'improve', 'should we do this', 'compare this to what we have'. Grounded by scripts/improveground.py in the lessons corpora and…
remem
Use when the user asks Codex to recall prior project context, save durable decisions or bug fixes, inspect remem memory health, or activate remem automatic memory hooks from the Codex plugin.
planning-with-scratchpad
Use when user explicitly requests planning with a scratchpad, or asks for persistent tracking of a complex task that the human can browse visually. Backs planning files with the scratch CLI so a pad's files are registered and viewable.