context-brain
Skill114 starsChanged 5 months agoMITView on GitHub
Build a reusable context pack (a "brain") so Claude reasons from the user's role, audience, voice, and standards every time, instead of generic output. Use when setting up a Project/workspace, or when output is on-topic but off-voice. Works for any role or domain.
1 other project uses a version of this.
Adopted by
Most recently taken up by
- ivanhapaz/marketing-context-braintheir own version, first seen Jul 15, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- ivanhapaz/marketing-context-brain★ 5same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository6 other files it gives its agents.
- connector-setupSkill
- orchestratorSkill
- pipeline-builderSkill
- prompt-architectSkill
- skill-authorSkill
- work-engineSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 3 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 34
Write a 2–3 line standing brief that loads on every chat: who we are, who we talk to, and the voice.
- Distinctive: line 38
Run one real task the user would actually ship, using the brain.
- Distinctive: line 47
Never assume the domain — build from the interview.
≈ “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
Context Brain
Out of the box, Claude writes well but generically. A context brain is a small, curated set of documents that make every future prompt start from your world. The skill is to curate, not dump — Claude reasons better from five sharp documents than fifty noisy ones.
Role-agnostic: an HR team's brain holds policy and tone; a designer's holds principles and references; a sales team's holds positioning and objection handling. Ask what theirs is.
Step 1 — Interview
Ask the user:
- What do you do, and for whom? Their role, their org/team, and who their work is for (the audience or stakeholder).
- What does "on-voice / correct" mean to you? How good output sounds or is judged in their world.
- What few references define it? Their best past work, a style/policy doc, positioning — whatever a new hire would need to be trusted. Aim for ~3–5 sources, not a folder of fifty.
- What should it never do? Words, claims, tones, or moves to avoid.
Step 2 — Build the four core docs
Produce these as clean, short markdown files the user can drop into a Project:
positioning.md— who they are, what they do, why it matters. A few sentences.voice-and-standards.md— how their work should sound/look/read, with do-and-don't pairs. Concrete, not adjectives.reference-examples.md— 1–3 short samples of "good", lightly annotated with why they're good.never-do.md— the short list of hard nos.
Draft from the interview; mark anything you inferred so the user can correct it.
Step 3 — Custom instructions
Write a 2–3 line standing brief that loads on every chat: who we are, who we talk to, and the voice. This is the always-on summary; the four docs are the depth behind it.
Step 4 — Test, then prune
Run one real task the user would actually ship, using the brain. If it's off, the knowledge is wrong, not the prompt — fix the docs. Then prune: remove anything that didn't change the output. Tighter context, sharper results.
Hand-off
Tell the user where to put these (a Claude Project, or a context/ folder they paste in). Once it exists, the prompt-architect skill can reference it instead of re-stating context every time.
Rules
- Curate, don't dump. Five sharp docs beat fifty.
- Never assume the domain — build from the interview.
- Show, don't tell: examples and do/don't pairs over adjectives.
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.