arthur-mcp
CLAUDE.md2 starsChanged 4 months agoMITView on GitHub
Full shared project memory (protocol, documentation gate, specialist index) lives in AGENTS.md — read it at session start: Arthur MCP — self-hostable control plane for building/operating MCP servers from APIs, databases, prompts, resources, and internal tools. There is no lint script configured (eslint is a devDependency but no eslint.config.*/lint script exists yet); Prettier formatting is defined in .prettierrc. Backend, frontend, and permission-gate conventions are split into .claude/rules/ (auto-loaded only for matching paths) instead of duplicated here — see .claude/rules/backend.md, .claude/rules/frontend.md,…
- Installs packages
In this repository3 other files it gives its agents.
- AGENTS.mdAGENTS.md
- add-permissionSkill
- session-handoffSkill
The file
We split this CLAUDE.md into the rules it gives an agent
1 rule. About 0 are common, 0 are distinctive, and 1 is 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.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 38
Every new page, tab, API endpoint, integration, credential surface, or settings panel must make an explicit permission decision (reuse or …
≈ “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
CLAUDE.md
Full shared project memory (protocol, documentation gate, specialist index) lives in AGENTS.md — read it at session start:
@AGENTS.md
Quick reference
Arthur MCP — self-hostable control plane for building/operating MCP servers from APIs, databases, prompts, resources, and internal tools.
- Monorepo: repo root = React 18 + Vite + TypeScript frontend;
api/= NestJS backend. - Persistence: SQLite by default, MongoDB optional via
DATABASEenv var (repository pattern with injection tokens). - Backend domains:
api/src/dynamic-mcp/(MCP tool generation/execution),api/src/swagger/(OpenAPI/Postman import),api/src/auth/,api/src/prompts/,api/src/secrets/,api/src/settings/,api/src/users/,api/src/roles/,api/src/observability/. - Frontend:
src/pages/,src/components/,src/context/,src/i18n.ts+src/locales/,src/theme/.
Commands
| Task | Command |
| Install deps (frontend/backend) | npm install / npm install --prefix api |
| Dev (both) | npm run dev |
| Dev frontend / backend only | npm run start / npm run start:dev --prefix api |
| Build frontend / backend | npm run build / npm run build --prefix api |
| Type-check frontend | npm run type-check |
| Test frontend | npm test |
| Test backend / e2e | npm test --prefix api / npm run test:e2e --prefix api |
| Backend coverage gate | npm run test:cov --prefix api -- --runInBand |
There is no lint script configured (eslint is a devDependency but no eslint.config.*/lint script exists yet); Prettier formatting is defined in .prettierrc.
Topic-scoped conventions
Backend, frontend, and permission-gate conventions are split into .claude/rules/ (auto-loaded only for matching paths) instead of duplicated here — see .claude/rules/backend.md, .claude/rules/frontend.md, and .claude/rules/permissions.md.
Non-negotiables (always apply)
- Database schema/data-shape changes go through migrations — never
synchronize, manual DB edits, or implicit startup sync. - Every new page, tab, API endpoint, integration, credential surface, or settings panel must make an explicit permission decision (reuse or add end-to-end) — see
.claude/rules/permissions.md. - Documentation is part of the deliverable: update
AGENTS.md,docs/ROADMAP.md,docs/HANDOFF.md,docs/ENTITIES.md,docs/DESIGN_PATTERNS.md, ordocs/FLOWS.mdwhen the area they cover changes. - Project docs, code identifiers, and translation keys stay in English; only
src/locales/<locale>/values are translated.
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.

