weclaws AGENTS.md

Repository-wide instructions for WeClaws, a multi-user software control system. They define its architecture, including SQLite as the stored source of truth, a supervisor that manages processes, and a web interface that records requested actions.

In plain words
What is it for?
Guiding work across the web app, supervisor, database package, shared contracts, authentication, bot processes, and event streaming.
Why use it?
They keep changes aligned with the system's ownership boundaries and prevent unsafe handling of processes, paths, secrets, and runtime data.

Instructions file for CodexOpenCode

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add instructions/yokingma/weclaws/agents-md
Clone the repo
git clone --depth 1 https://github.com/yokingma/weclaws

Made for: Codex, OpenCode.

Per session 696 This file is loaded in full into every session.
When invoked 696 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00696 $0.00696
Opus 5 $0.00348 $0.00348
Sonnet 5 $0.00139 $0.00139
Haiku 4.5 $0.00070 $0.00070

Measured 2d ago against content hash a0d4653f1b0d, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

weclaws AGENTS.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

AGENTS.md · 75 lines

What it actually says

AGENTS.md

This file sets repository-wide rules for coding agents in WeClaws.

Read first: README.md and docs/manuals/README.md.

Working goal: Small, accurate, verifiable changes aligned with the current architecture.

Priority:

  1. User task
  2. This file + relevant workspace PATTERNS.md
  3. README.md and docs/manuals/*
  4. Existing code patterns

Core rules:

  • Supervisor owns runtime; web writes durable intents only.
  • SQLite is the source of truth; persistence logic lives in packages/db.
  • Use FastAgent only through external contracts.
  • Derive bot paths through shared resolvers; do not persist host-specific paths or commit runtime data, secrets, or artifacts.
  • Default scope: single-machine, multi-user control plane, one bot per child process, SQLite, Better Auth email/password, per-bot SSE.

Workspace rules:

  • apps/web: no local process management; intents first, supervisor reconciles.
  • apps/supervisor: consume runtime JSONL and runtime state; child env must be allowlisted.
  • packages/db: narrow semantic APIs; fresh SQLite schema must work; keep migrations in sync.
  • packages/shared: only stable cross-workspace contracts; no compatibility layers unless explicitly required.

Development norms:

  • Prefer test-first for new behavior and bug fixes.
  • Keep diffs small and focused.
  • Use the smallest verification set that covers the change.
  • Prefer existing patterns and shared contracts.
  • Keep rules concise.

Do not:

  • Let web own runtime.
  • Treat in-memory state as cross-process truth.
  • Add retired-behavior compatibility unless required.
  • Present scratch notes as user-facing docs.
  • Claim tests, builds, or smoke checks passed unless actually run.

Common commands:

  • pnpm install
  • pnpm build
  • pnpm test
  • pnpm typecheck
  • pnpm lint
  • pnpm db:generate
  • pnpm db:migrate
  • pnpm --filter @weclaws/web test
  • pnpm --filter @weclaws/supervisor test
  • pnpm --filter @weclaws/db test
  • pnpm --filter @weclaws/shared typecheck

Environment and deployment:

  • Node 20.18.1 is recommended.
  • Root .env is for local dev; infra/compose/.env is for Compose.
  • FASTAGENT_SANDBOX_MODE supports remote and disabled; remote uses owner-specific pool creds.
  • The default Compose path is the repo-local three-image topology.
  • Read the Docker runbook before Compose changes.

Verification:

  • Run the smallest matching verification set.
  • UI/API changes: web test / typecheck, build only when needed.
  • Runtime/supervisor changes: supervisor test / typecheck.
  • DB changes: db test / typecheck and migrations when required.
  • Shared contract changes: affected workspace tests.
  • Run pnpm test:fastagent-contract only when the required env exists.
  • Report precisely: passed / not run / blocked.

Documentation:

  • Update workspace CHANGELOG.md and PATTERNS.md when executable code or behavior changes.
  • Keep current facts in README.md and docs/manuals/.
Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 75 lines · 696 tokens per session scan A a0d4653f1b0d

Subscribe to this mod's changes

weclaws AGENTS.md is an instructions file published in the GitHub repository yokingma/weclaws (38 stars, last pushed 3mo ago), licensed MIT. It adds 696 tokens to every session, about $0.0035 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.