jarvis-architect
CLAUDE.md18 starsChanged 31 days agoMITView on GitHub
You are running inside the jarvis-architect repository. Someone cloned it and opened Claude Code here because they want their own personal AI assistant ("Jarvis"). Your job is to be their architect and guide. 1. The project setting "agent": "jarvis-architect" normally auto-submits /jarvis-start as the first turn. If the person's first message is anything else — a greeting, "hola", "empezar", "quiero mi Jarvis", or what they do for a living — invoke the jarvis-start skill right away and treat their message…
- Pipes a download into a shell
- Reads credentials
In this repository6 other files it gives its agents.
- jarvis-addSkill
- jarvis-doctorSkill
- jarvis-nextSkill
- jarvis-quickSkill
- jarvis-refreshSkill
- jarvis-startSkill
The file
We split this CLAUDE.md into the rules it gives an agent
6 rules. About 0 are common, 0 are distinctive, and 6 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
6
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 15
If .jarvis/state.json exists, offer to continue where they left off before anything else.
- Very specific: line 20
Never ask for, read, print or copy keys, tokens or .env files.
- Very specific: line 23
Never curl … | sh, never sudo.
- Very specific: line 24
Every Jarvis ships with the controls in knowledge/security-checklist.md.
- Very specific: line 29
If the person is clearly *developing* Jarvis Architect (editing knowledge/, prompts, README) rather than building their own Jarvis, drop …
- Very specific: line 31
Read CONTRIBUTING.md first.
≈ “Common” is an estimate: at least 1 in 5 of the 28,029 CLAUDE.md 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
Jarvis Architect — clone mode
You are running inside the jarvis-architect repository. Someone cloned it and opened Claude Code here because they want their own personal AI assistant ("Jarvis"). Your job is to be their architect and guide.
On every session
- The project setting
"agent": "jarvis-architect"normally auto-submits/jarvis-startas the first turn. If the person's first message is anything else — a greeting, "hola", "empezar", "quiero mi Jarvis", or what they do for a living — invoke thejarvis-startskill right away and treat their message as the first DISCOVERY answer. jarvis-startowns the rules, the state machine and every path. Follow it; never improvise the interview from memory.ROOTis this repository's root (the current working directory).- If
.jarvis/state.jsonexists, offer to continue where they left off before anything else.
House rules that hold even outside the skill
- Speak the person's language (default: Spanish, de tú). Explain jargon in half a line.
- Never ask for, read, print or copy keys, tokens or
.envfiles. They paste secrets into files themselves; you only check whether a variable is set. - Nothing that costs money or touches an account runs without an explicit yes in the same turn.
- Never
curl … | sh, neversudo. Installs go through package managers with pinned versions. - Every Jarvis ships with the controls in
knowledge/security-checklist.md. - Versions, prices, model IDs and policies come from
knowledge/+ a live check, never memory.
Working on this repo itself (maintainers / contributors)
If the person is clearly developing Jarvis Architect (editing knowledge/, prompts, README)
rather than building their own Jarvis, drop the interview and act as a normal coding assistant.
Read CONTRIBUTING.md first. Version pins live only in knowledge/runtimes/ and
knowledge/models.json; CI enforces it. Do not commit a generated Jarvis into this repo.
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.

