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/zyqzyq/unfour/agents-mdgit clone --depth 1 https://github.com/zyqzyq/UnfourWrote 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/zyqzyq/unfour/agents-md)<a href="https://agentmods.dev/instructions/zyqzyq/unfour/agents-md"><img src="https://agentmods.dev/badge/instructions/zyqzyq/unfour/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 5.1 | $0.01528 | $0.01528 |
| Opus 5 | $0.00764 | $0.00764 |
| Sonnet 5 | $0.00306 | $0.00306 |
| Haiku 4.5 | $0.00153 | $0.00153 |
Grade A, and why
Unfour 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 yesterday.
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 — 168 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
Unfour is a lightweight IDE-style desktop developer tool built on Tauri 2, React, TypeScript, and Rust.
Product And Module Scope
- Primary v0.1 modules are API Client, SSH Terminal, Database, and Workspace.
workspace-coreowns shared frontend workspace state.workspace-localis the reserved frontend boundary for local workspace lifecycle, persistence, import/export, recent-workspace, and migration behavior.- Current release readiness belongs in
docs/release/*anddocs/testing/*, not in temporary progress logs.
Package Boundary Rules
packages/app-shellis limited to global layout composition, sidebar and module mount surfaces, top-level navigation wiring, route assembly, and cross-module container slots.- Do not add API request execution, SQL editing/execution, SSH session state,
feature mock data, or feature-specific large UI components to
packages/app-shell. - API Client, SSH Terminal, Database, and Workspace business state and business components must live in their owning packages or crates.
- Shared UI primitives must be reused from or added to
packages/ui. packages/uimust not contain feature-specific business logic.- Feature packages must not depend on
packages/app-shell. - Do not make unrelated cross-package changes. If a fix requires crossing a package boundary, explain why in the final report.
- Do not add new dependencies unless the task explicitly requires them and the reason is documented.
- Do not delete code with unconfirmed purpose for cosmetic reasons.
- New user-visible frontend UI copy must use the shared i18n provider/hook and locale keys instead of hardcoded strings. Do not implement parallel i18n systems inside individual packages.
Backend And Command-Bus Rules
- Frontend interaction lives in React/TypeScript; execution and security boundaries live in Rust.
apps/desktop/src-tauriis the Tauri adapter and composition layer. Backend capability logic belongs incrates/*.- Business actions for API, SSH, Database, Workspace, and future MCP/AI surfaces must route through the Rust command bus boundary.
- Tauri commands and MCP tools are adapters over the command bus, not places for duplicated domain logic.
- Cloud-bound local mutations must retain durable outbox intent regardless of login, active account, entitlement, pause, offline, or worker state; see Cloud Sync invariants.
- A local workspace has at most one durable Cloud Sync owner
(account_id, cloud_workspace_id); mutations and repair must resolve that owner and must never fan out across historical bindings. - MCP tool names, schemas, command names, command keys, and stable error codes must remain English and must not be localized. UI layers may map stable message keys to localized user-facing copy.
- Every persisted business record must carry
workspace_idunless it is truly global app configuration. - Passwords, private-key passphrases, API tokens, and database passwords must not be stored in SQLite plaintext. Persist only credential references.
authorization,cookie,proxy-authorization,x-api-key, andx-auth-tokenmust be redacted in logs, history, and local activity details.- Internal engineering, architecture, and agent documentation does not need multilingual versions unless a task explicitly asks for it.
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.
- yesterday Changed · +6 lines · +89 tokens per session daf4c0ac67dd
- 5d ago First seen · 162 lines · 1,439 tokens per session scan A 58db1f010d7a
Unfour AGENTS.md is an instructions file published in the GitHub repository zyqzyq/Unfour (2 stars, last pushed yesterday), licensed Apache-2.0. It adds 1,528 tokens to every session, about $0.0076 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 2026-08-31.
Other instructions, from other repositories
anarlog AGENTS.md
AGENTS.md instructions for fastrepl/anarlog, covering commands, pre-commit verification, guidelines, code style and cli tui command architecture.
anarlog CLAUDE.md
Claude Code instructions for fastrepl/anarlog, a project described as: Open source Granola AI Alternative.
openhuman AGENTS.md
AGENTS.md instructions for tinyhumansai/openhuman, covering openhuman, repository layout, runtime scope, ios client (experimental, non-shipping) and commands (from repo root).
terax-ai AGENTS.md
AGENTS.md instructions for crynta/terax-ai, a project described as: Lightweight (7MB) Terminal-first AI-native dev workspace.
phase CLAUDE.md
Claude Code instructions for phase-rs/phase, covering claude.md, project overview, design principles — read this first, when in doubt and agent team orchestration standards.
openhuman CLAUDE.md
Claude Code instructions for tinyhumansai/openhuman, a project described as: OpenHuman is an open source personal AI for Mac, Windows and Linux — local-first memory, agent orchestration, and deep research.