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.
npx agentmods add instructions/automagik-dev/genie/agents-mdgit clone --depth 1 https://github.com/automagik-dev/genieWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/instructions/automagik-dev/genie/agents-md)<a href="https://agentmods.dev/instructions/automagik-dev/genie/agents-md"><img src="https://agentmods.dev/badge/instructions/automagik-dev/genie/agents-md.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.01170 | $0.01170 |
| Opus 5 | $0.00585 | $0.00585 |
| Sonnet 5 | $0.00234 | $0.00234 |
| Haiku 4.5 | $0.00117 | $0.00117 |
Grade A, and why
genie 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 4d 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.
How it starts
The opening of the file, as written. The whole thing — 64 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Genie repository contract
This is the runtime-neutral contributor contract for Claude Code, Codex, and human contributors. Client-specific overlays may add syntax, but they must not contradict this file.
Validate changes
bun install --frozen-lockfile
bun run check
bun test path/to/file.test.ts
The full gate runs type checking, Biome, dead-code analysis, skill/wish/council linting, and tests. Tests use bun:test; fixtures belong under a temporary directory and must isolate GENIE_HOME when touching global state.
Architecture
src/genie.tsis the Commander CLI entry point.src/lib/v5/owns SQLite state. Per-repo.genie/genie.dbstores task state; global~/.genie/genie.dbstores Omni state. Never mix their path/schema modules.src/hooks/owns provider-neutral lifecycle policy plus Claude/Codex wire adapters.src/term-commands/ownsinit,context, MCP, Omni, task, and board commands.plugins/genie/is one shared plugin payload with sibling Claude and Codex manifests.skills/is shared runtime-neutral workflow guidance. Runtime mapping lives inplugins/genie/references/native-surfaces.md..genie/contains git-tracked wishes/brainstorms/index plus gitignored operational SQLite files.
Genie v5 is zero-daemon except for the explicitly launched genie omni serve bridge. Do not use telemetry presence as integration health.
Engineering rules
- KISS is a release gate, not a preference. Start with the simplest complete design that satisfies current user stories. Caches, deltas, sharding, background coordination, configurable policy, and other stateful machinery require a present contractual need or measured threshold; hypothetical future scale is not evidence. Prefer bounding data and separating history from current state before adding synchronization protocols.
- Define type and error boundaries before implementation.
- Preserve user-owned config and unrelated dirty-worktree changes.
- Config migrations are narrow, backup-first, idempotent, and covered by fixtures.
- Every new CLI surface tests success, error exit code, stderr, and idempotency.
- Shared skills use roles and native delegation language, never a hardcoded client tool name.
- Subagents share a workspace unless the client explicitly guarantees otherwise. Parallel writers must have disjoint file ownership or dedicated worktrees; otherwise sequence them. Shared-workspace subagents never mutate repo-level git state (no
checkout/switch/reset/stash/rebase) — only the orchestrator moves HEAD; work needing repo-level mutation gets an isolated worktree arranged by the orchestrator (client-provided worktrees or explicitgit worktree addplumbing) or gets sequenced.git worktree add/remove/pruneon snapshot/lane paths is orchestrator-side plumbing and permitted. Genie task claims own shared-workspace scope; the orchestrator arranges worktree isolation. - Reviewer and engineer are different roles. Never accept self-review as independent evidence.
- Codex agents inherit the active model; do not hardcode unstable model identifiers.
- Hook trust and workspace trust remain explicit user decisions.
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.
- 4d ago First seen · 64 lines · 1,170 tokens per session scan A bc2d4c8ef05c
genie AGENTS.md is an instructions file published in the GitHub repository automagik-dev/genie (334 stars, last pushed 3d ago), licensed MIT. It adds 1,170 tokens to every session, about $0.0059 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.
Other instructions, from other repositories
codex-autoresearch AGENTS.md
Instructions for TheGreenCedar/codex-autoresearch, covering agents.md, purpose and scope, canonical sources, stable product boundaries and source and package boundaries.
aider-desk AGENTS.md
AGENTS.md instructions for hotovo/aider-desk, covering agents.md, common commands, development, type checking and linting and formatting.
OwnPilot AGENTS.md
Instructions for ownpilot/OwnPilot, covering ownpilot, architecture, key patterns, commands and tech stack.
nightshift AGENTS.md
Instructions for orwa-mahmoud/nightshift: AI-assisted contributions are welcome. The same correctness, security, licensing, testing, and quality standards apply regardless of which tools are used.
SeekMoney-ai CLAUDE.md
Instructions for zykooooooooo/SeekMoney-ai, covering claude.md, project overview, development commands, development and production.
simplicio-loop AGENTS.md
Instructions for wesleysimplicio/simplicio-loop, covering agents.md — simplicio-loop, simplicio ecosystem contract (canonical), worker startup and centralized artifacts (mandatory), full-stack boundaries and what to load.