longread
Skill16 starsChanged 5 months agoMITView on GitHub
Write an analytical longform piece — a 4-phase pipeline from raw idea to finished text, with research-backed facts and a devil's-advocate pass before publishing. Invoke /longread <topic>. Triggers — "write a longread about X", "write an analytical breakdown of Y", "write an expert post", "turn this into an article".
In this repository9 other files it gives its agents.
- goalSkill
- konsiliumSkill
- pre-pushSkill
- proofSkill
- ship-secureSkill
- system-healthSkill
- teachSkill
- vetSkill
- workflow-upgradeSkill
The file
We split this skill into the rules it gives an agent
7 rules. About 0 are common, 6 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈6
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 31
Show the structure and confirm: before Phase 2.
- Very specific: line 39
Verification rule: every concrete number needs a source.
- Distinctive: line 42
Write in a plain, conversational voice, not an academic one.
- Distinctive: line 45
No filler openers ("It should be noted that…", "Thus…").
- Distinctive: line 46
No AI-aesthetic ("In conclusion, it is worth noting…").
- Distinctive: line 55
Write the hook last (once you know what actually grabs after the body is done).
- Distinctive: line 70
If relevant: a note on what illustration/image type fits.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Longread Writer
A pipeline for analytical writing. Structure first, facts second, voice third, self-critique last. Works for any subject — adapt the sources to the domain.
Output formats
| Format | Length | Where it goes |
| Short expert post | 300–600 words | social / a channel |
| Longform post | 800–1500 words | blog / newsletter |
| Deep analytical breakdown | 1500–3000 words | a standalone piece |
| Case study | 1000–2000 words | portfolio / blog |
Pipeline (4 phases)
Phase 1 — Cheap: angle & structure (don't write yet)
- Clarify if unclear: format? audience? any data/insights that must be included?
- State the central thesis in one sentence — what the reader walks away with.
- Propose a structure (3–5 blocks):
Hook: [a provocative fact / question / contradiction] Block 1: [name] Block 2: [name] … Conclusion / CTA: [what the reader should do] - Show the structure and confirm before Phase 2.
Phase 2 — Research: gather facts
Without specifics, the text is empty. For each block:
- Find 1–2 key facts/figures.
- Find 1 concrete example or case.
- Note the source (URL / dataset / first-hand).
Verification rule: every concrete number needs a source. Can't find one → write [need figure] and move on. Never invent a number.
Phase 3 — Write: section by section
Write in a plain, conversational voice, not an academic one.
- 2–4 sentences per thought, no more.
- Concrete: numbers, names, dates — not "many", "often", "significantly".
- No filler openers ("It should be noted that…", "Thus…").
- No AI-aesthetic ("In conclusion, it is worth noting…").
Each block's shape:
Claim (1 sentence — the block's point)
Evidence (fact / figure / example)
Why it matters
[transition to the next block]Write the hook last (once you know what actually grabs after the body is done). Write section by section; show intermediate drafts if approval is useful.
Phase 4 — Devil's advocate: pre-publish check
Ask yourself 5 questions:
- "What if I'm wrong?" — is there a counterargument to the thesis? How do you handle it?
- "Where do I generalize without data?" — find "many", "as a rule", "obviously" — replace with specifics or cut.
- "What won't the reader get?" — any terms unexplained for a non-specialist?
- "What's stale?" — any data that may be outdated (>6 months)?
- "Why read to the end?" — is there an interest dip in the middle?
Make targeted edits, don't rewrite everything.
Final
- Title (for a post — first line as a bold headline).
- The text in the chosen format.
- If relevant: a note on what illustration/image type fits.
Anti-patterns
- ❌ Writing without a structure — always Phase 1 first.
- ❌ Facts without a source — better
[need figure]than an invented number. - ❌ Academic style ("this work is dedicated to…").
- ❌ A longread of generalities — if there's no concrete data, say so before writing.
- ❌ Skipping the devil's-advocate pass — that's where the weak spots are.
Examples of a strong hook
- "A prediction market paid traders $2.3M in one day. Half of it went to 12 wallets."
- "Everyone knows BTC correlates with macro. Nobody says how much — by timeframe it runs from 0.1 to 0.9."
- A specific number + a specific stake beats any abstract claim.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

