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/mdrideout/junjo/agents-mdgit clone --depth 1 https://github.com/mdrideout/junjoWhat 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 | $0.01556 | $0.01556 |
| Opus 5 | $0.00778 | $0.00778 |
| Sonnet 5 | $0.00311 | $0.00311 |
| Haiku 4.5 | $0.00156 | $0.00156 |
Grade A, and why
junjo 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.
How it starts
The opening of the file, as written. The whole thing — 113 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Junjo Platform AGENTS.md
Junjo is a multi-component platform repository. Read this file first, then read
the nearest scoped AGENTS.md before changing a component.
Developer philosophy
- Be grug brained: prefer simple, explicit code and contracts.
- Everything is greenfield. Intentional breaking changes are allowed, but they must be documented and coordinated across every affected component.
- Do complete, well-architected work. Do not add compatibility fallbacks or abstractions that hide ownership.
- Follow single responsibility and separation of concerns.
- Ground plans and reviews in current code and accepted ADRs.
- Avoid scope creep and preserve unrelated user work.
- Do not engage in scope creep. Do not take liberties to refactor or change things that do not need to change beyond the requested implementations and ideas. Keep existing user-interfaces, styles, contracts, integrations, system -> system mechanics as they are unless it's required to change them as part of new feature implementation. Keep changes necessary and required. As much as needed, as little as possible.
- Do not set arbitrary contraints, budgets, caps, limitations. Do not truncate content. Do not set timeouts. Do not make assumptions about how much we can handle. Allow us to run into the exceptions when resources, time lengths, or capacity are exceeded. We will only create constraints as we encounter real exceptions caused by a real repeatable documented problem or limit.
Scope and Complexity
Simplicity is valuable. The more complex we make the systems, the more edge cases for issues there are, the more testing we need to do, the more brittle the system becomes, and the higher the blast radius of all future changes. This is why we avoid scope increases and complexity and avoid these problems.
We do not need to mitigate or handle all exceptions at all costs. Allowing exceptions to happen especially for very rare or transient circumstances is better than creating complexity and architecture changes to accommodate a very rare or temporary situations.
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.
- 2d ago First seen · 113 lines · 1,556 tokens per session scan A a7f1e7d5c0c7
junjo AGENTS.md is an instructions file published in the GitHub repository mdrideout/junjo (20 stars, last pushed 2d ago), licensed Apache-2.0. It adds 1,556 tokens to every session, about $0.0078 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
watchtower-ai CLAUDE.md
Instructions for gugastork/watchtower-ai, covering claude.md, about this repository, content standards, repository structure and writing style.
llm-agnostic-icm-framework AGENTS.md
Instructions for alejandro-ventures/llm-agnostic-icm-framework, covering agents.md — workspace router (layer 0), how to use this workspace, the one hard boundary, instruction precedence and routing table.
llm-agnostic-icm-framework copilot-instructions.md
Instructions for alejandro-ventures/llm-agnostic-icm-framework: The instructions for this workspace live in /AGENTS.md. Read it first and follow it.
llm-agnostic-icm-framework CLAUDE.md
Instructions for alejandro-ventures/llm-agnostic-icm-framework: The instructions for this workspace live in /AGENTS.md. Read it first and follow it.
AutoAgents AGENTS.md
Instructions for liquidos-ai/AutoAgents, covering autoagents repository guidelines, project structure & module organization, code quality, naming conventions and error handling.
claude-code-karma CLAUDE.md
Instructions for JayantDevkar/claude-code-karma, covering claude.md, project overview, quick start, clone (no submodule init needed — this is a monorepo) and start api (terminal 1).