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/adarshp14/relay/agents-mdgit clone --depth 1 https://github.com/adarshp14/relayWhat 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 | $0.00600 | $0.00600 |
| Opus 5 | $0.00300 | $0.00300 |
| Sonnet 5 | $0.00120 | $0.00120 |
| Haiku 4.5 | $0.00060 | $0.00060 |
Grade A, and why
relay 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 — 63 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
Task Completion Requirements
- Use Effect Vitest for tests.
- Run targeted tests with
vitest run ...when working on a scoped area. - The root/package
bun run testscripts are allowed because they delegate to Vitest. - NEVER run
bun test. - For code changes, run the narrowest useful verification before handing back.
- For broad or merge-ready changes, the full gates are
bun run format:check,bun run lint,bun run typecheck, andbun run test.
Attribution
Do not add any AI assistant, Claude, Anthropic, or Co-Authored-By attribution/trailers to commits, commit messages, PRs, or generated files.
Pull request titles and descriptions are going to a public GitHub repo, so avoid using specific names or internal info unless explicitly stated to.
Collaboration Notes
The user uses speech to text occasionally, so if sentences are weird or words are not right, infer the likely intent and ask only when needed.
Code is very cheap to write. Do not give time estimates; with agents, code is practically instant to generate. Unless stated otherwise, time to implement is not a blocker.
Reference Repos
Repos in .reference, such as Effect and effect-atom, are available for
patterns. If given a Git URL for reference, clone it into .reference and
inspect it there. Make sure to pull the latest changes from the reference repo
before using it.
Engineering Priorities
- Prefer correctness and predictable behavior over short-term convenience.
- Preserve runtime behavior when changing lint, typing, or test structure.
- Keep package boundaries clear; use public package exports instead of relative imports across package roots.
- Extract shared logic only when the shared behavior is real and local patterns support it. Avoid broad generic abstractions for one-off duplication.
Package Roles
packages/core/sdk: relay core contracts, plugin wiring, scopes, sources, secrets, policies, and test fixtures.packages/core/storage-*: storage adapters and storage test support.packages/plugins/*: protocol and provider plugins. Plugin-specific runtime, React, API, and testing helpers should live with the owning plugin.packages/react: shared React UI and atom/client integration.packages/hosts/mcp: MCP host surface for exposing Relay through MCP.packages/kernel/*: execution runtimes and code execution substrate.apps/local,apps/cloud,apps/cli, andapps/desktop: product entry points that compose the packages.
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 First seen · 63 lines · 600 tokens per session scan A cf1d5ffbc4d3
relay AGENTS.md is an instructions file published in the GitHub repository adarshp14/relay (0 stars, last pushed 2mo ago), licensed MIT. It adds 600 tokens to every session, about $0.0030 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
api-gateway CLAUDE.md
Instructions for membrane/api-gateway, covering claude.md, working principles, 1. think before coding, 2. simplicity first and 3. surgical changes.
LAP CLAUDE.md
Instructions for Lap-Platform/LAP, covering lap - lean api platform, what this is, quick reference, install and run tests.
webiny-js AGENTS.md
AGENTS.md instructions for webiny/webiny-js, covering exploration, persist learnings, code, building and testing.
webiny-js CLAUDE.md
Claude Code instructions for webiny/webiny-js, covering context, claude project guidelines, reading code, development commands and install dependencies (suppress all output – it's noise).
webiny-js copilot-instructions.md
Copilot instructions for webiny/webiny-js, a project described as: Open-source, self-hosted CMS platform on AWS serverless (Lambda, DynamoDB, S3). TypeScript framework with multi-tenancy, lifecycle hooks, GraphQL API, and AI-assisted development via MCP server. Built for developers at large organizations.
caplets AGENTS.md
Instructions for spiritledsoftware/caplets, covering agents.md, commands, package map, agent skills and issue tracker.