Skip to content
Agentmart
Submit a tool

learn

Research an unfamiliar domain from primary sources into notes or a publish-ready draft, keeping contradictions visible.

Skill by therealtinhtute

IndexedMIT★ 5
Write a review

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/therealtinhtute-learn

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Research an unfamiliar domain from primary sources into notes or a publish-ready draft, keeping contradictions visible. Use for deep dives. Not for one-URL lookups (use read).

Use cases: research, productivity

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md and code. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/learn

What's inside

Skill name
learn
Description
Research an unfamiliar domain from primary sources into notes or a publish-ready draft, keeping contradictions visible. Use for deep dives. Not for one-URL lookups (use read).
License
MIT
Compatibility
Designed for Claude Code
Instructions
42 lines in SKILL.md · view on GitHub

Files (2)

  • NOTICE.md
  • SKILL.md

SKILL.md (first lines)

Support the user's thinking; do not replace it.

## Mode

| Mode | Entry | Exit |
|---|---|---|
| Quick Reference (default when unsure) | Collect | Digest: notes only |
| Deep Research | Collect | Self-review: publish-ready draft |
| Write to Learn (materials supplied) | Outline | Self-review |
| Canonical Article ("one page is enough") | Collect | Self-review, plus worked examples, common mistakes and 3–5 Further Reading picks |

Ask only when two modes would change the deliverable.

## Phases

1. **Collect** — primary sources only: papers, official docs and blogs, builders' posts, reference repos. Three ordered steps:
   - Discover: `read` search maps the landscape, then 2–3 deep searches on the best sub-topics. Output is a URL list.
   - Fetch: every URL through `read`, which owns the cascade, privacy and paywall checks.
   - File: pass `read` the source directory (default: scratchpad `learn/<topic>/sources/`); move, never refetch.
   Target 5–10 sources for a post, 15–20 for a survey.
2. **Digest** — read each source fully; keep a claim only if it recurs in the source, predicts new cases, or is specific to that source rather than generic wisdom. Summarizing is not digesting.
3. **Outline** — every section names its sources; a section without one is cut or sent back to Collect.
4. **Fill in** — section by section; a section that resists writing goes back to Digest for that sub-topic only.
5. **Refine** — edits only: cut repetition, flag broken flow and unexplained terms, then run `write` to strip AI patterns.
6. **Self-review** — the user reads the draft linearly, twice; fix what they mark.

## Hard Rules

- No Fill in before every outline section has a source.
- Contradictions stay visible: state both positions and each one's evidence.
- Every factual claim cites a fetched source URL or file.
- Stop at publish readiness; never post, upload or publish unless asked.

## Output

The mode's exit artifact, its saved path, and a Sources list (URL, tier, one-line why). Prose for conclusions, a diagram only when relationships carry the point.

Defer to: `read` for every search and fetch; `write` for prose polish; `think` when the question is a decision, not a domain.

From therealtinhtute/mono-harness, MIT.

More from therealtinhtute/mono-harness

Signals

Stars
5
Forks
1
Open issues
1
Last commit (this folder)
Oct 23, 2026 (5 days ago)
Last release
Oct 23, 2026 (6 days ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from therealtinhtute

Cite this page

learn skill. Agent Mart. https://agentmart.dev/skills/therealtinhtute-learn (last updated 2026-10-24)