Ephesus
CLAUDE.md2 starsChanged 3 months agoMITView on GitHub
Ephesus is a multi-agent harness desktop app: a "company of agents" governed by a human Architect. It was built in Electron + React + TypeScript + Pixi.js + xterm.js + node-pty, and is being rewritten in Rust, window included (ADR-0041, 2026-10-19). The TypeScript tree is frozen as the behavioural reference until parity. The company's primary standing mission is to improve itself — every session here either builds the system or improves how it is built, and both flow through the…
1 other project uses a version of this.
Adopted by
Most recently taken up by
- mertefesensoy/Ephesusword for word, first seen Aug 26, 2026
Other versions1 project uses a version of this.
Word for word · 1
- mertefesensoy/Ephesus★ 2100% same
Edited · 0
None found.
Wrote their own version · 0
None found.
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 repository7 other files it gives its agents.
- build-packageSkill
- doc-syncSkill
- ephesus-engineerSkill
- goalSkill
- improveSkill
- milestone-reviewSkill
- researchSkill
The file
We split this CLAUDE.md into the rules it gives an agent
1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
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 26
Read README.md first for the subsystem map (Artemis, Hermes, Agora, Library, Odeon, Herald, Harbor, Watch, Terraces, Gymnasium).
≈ “Common” is an estimate: at least 1 in 5 of the 26,980 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
Ephesus — project memory for Claude Code sessions
Ephesus is a multi-agent harness desktop app: a "company of agents" governed by a human Architect. It was built in Electron + React + TypeScript + Pixi.js + xterm.js + node-pty, and is being rewritten in Rust, window included (ADR-0041, 2026-10-19). The TypeScript tree is frozen as the behavioural reference until parity. The company's primary standing mission is to improve itself — every session here either builds the system or improves how it is built, and both flow through the same documentation and gates.
What kind of session is this?
- Implementing the system → follow
BUILD-PROMPT.mdexactly. It defines your role, reading protocol, work packages, and report format. Resume from the Resume here block at the top ofdocs/PROGRESS.md— it names the next step and what waits on the Architect. As of 2026-10-19 that is the Rust rewrite,docs/RUST-PLAN.md(R0 → R14), which runs before M9. M9 is still planned and not authorised to start. - Improving the system/process → follow the Gymnasium loop:
docs/adr/ADR-0015-gymnasium-self-improvement.md. Improvement work is proposed, gated, and measured — never freelanced. - Editing documentation → respect precedence (below) and cross-reference rules: a requirement lives in one place and is linked from everywhere else, never restated.
Source of truth and precedence
Read README.md first for the subsystem map (Artemis, Hermes, Agora, Library, Odeon,
Herald, Harbor, Watch, Terraces, Gymnasium). When documents seem to disagree:
SDD > ADR > SRS > README for how; SRS > SDD for what. ADRs are append-only —
never edit an accepted ADR; supersede it.
| Need | File |
| Requirements, use cases, acceptance tests | docs/srs/SRS.md |
| Why decisions were made | docs/adr/ |
| Architecture, data models, IPC, sequences | docs/sdd/SDD.md |
| Coding/security rules, Definition of Done | docs/ENGINEERING-STANDARDS.md |
| Trust boundaries, what each control stops, residual risks | docs/THREAT-MODEL.md |
| What tests are owed | docs/TEST-STRATEGY.md |
| Build order and current state | docs/IMPLEMENTATION.md + docs/PROGRESS.md (its Resume here block; closed milestones in docs/progress/) |
| Claude Code automation in this repo | docs/AUTOMATION.md |
Commands
The Rust workspace (from R0.1; set CARGO_TARGET_DIR outside OneDrive first):
cargo xtask gate # fmt, clippy -D warnings, test, deny, invariants, deps (RUST-PLAN §4)The frozen TypeScript tree (its CI still guards main until R14):
npm run dev # Electron app with hot reload
npm run typecheck # node + preload + web projects; must be green at every commit
npm run lint # ESLint + Prettier, zero warnings
npm test # VitestNon-negotiable invariants (digest — full list in BUILD-PROMPT.md §3)
Strict TS, no any past validators · renderer never touches Node/fs · atomic writes
(temp+rename) for shared files · only main commits to the Agora · append-only stays
append-only · secrets write-only, env-injected · every degradation visible in UI ·
prompt text lives in prompts/, never in code · schema'd files carry schemaVersion ·
no new dependencies without a decision memo · UI values come only from design tokens.
Conventions
Conventional Commits, subject ≤ 72 chars. Branches feature/<topic>, fix/<topic>,
agent/<name>/<topic>. Every PR carries evidence (screenshot / terminal capture /
test output). One work package per PR. Do not create PRs unless asked.
Project skills available here
/ephesus-engineer (the operating protocol the others are executed to: when to
consult the Architect, the invariants, the evidence tiers, honest verification
reporting) · /goal (drive the current milestone run: authorship + branch policy +
package loop) · /build-package (execute next BUILD-PROMPT work package) ·
/milestone-review (verify exit criteria, update PROGRESS) · /doc-sync
(code↔docs drift check) · /improve (file a Gymnasium improvement proposal) ·
/research (run a Stoa research cycle over the watchlist).
See docs/AUTOMATION.md.
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.