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/morpho-org/sdks/agents-mdgit clone --depth 1 https://github.com/morpho-org/sdksWhat 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 | $0.08097 | $0.08097 |
| Opus 5 | $0.04049 | $0.04049 |
| Sonnet 5 | $0.01619 | $0.01619 |
| Haiku 4.5 | $0.00810 | $0.00810 |
Grade A, and why
sdks 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 — 266 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Engineering Rules
Mission, scope, and values: see
MISSION.md. The rules below are how we build it.
CLAUDE.md is a symlink to AGENTS.md at every level — same content, two filenames so any agent finds it. Per-package AGENTS.md files refine these rules for a specific package; they may add detail but must not contradict this file. When in doubt, the root wins.
Every PR is measured against the rules below. A change that violates an architectural principle doesn't land — we question the change before we question the principle.
Enforcement note. Some rules below are enforced by tooling today (Biome formatter, fork harness in
@morpho-org/test, Changesets generation). Most are review-time conventions that humans and reviewing agents apply: JSDoc on every export, layered-import bans, the §2 forbidden-patterns list (Biome'snoExplicitAnyis warn-level,noParameterAssignis disabled, and there's no rule banningas unknown as/@ts-ignore/ async-in-actions / framework imports / mocked viem clients on RPC paths), changeset-gates-CI, full coverage thresholds. Where a rule isn't backed by an automated check, treat it as binding regardless — wiring CI gates is tracked separately.
Review personas. The review-time conventions above are applied at PR review by specialized personas under
.agents/pr-review-engine/agents/, invoked by the/pr-review-{ci,gh,local}slash commands. See §10 for the full inventory and the CI/release rules they anchor. When a rule below changes, the matching persona's bullet must change with it — the backlinks on each section name the personas to update.
1. Architecture
The three pillars: layering, modularity, testability. Everything else (types, API, tests, docs) flows from getting these right.
Layering: one direction, no exceptions
Client → Entity → Action. Dependencies point one way. Lower layers never reach back into higher ones.
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 · 266 lines · 8,097 tokens per session scan A b2ae7e1c8325
sdks AGENTS.md is an instructions file published in the GitHub repository morpho-org/sdks (40 stars, last pushed yesterday), licensed MIT. It adds 8,097 tokens to every session, about $0.0405 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 2026-09-01.
Other instructions, from other repositories
skills docs.instructions.md
Instructions for apollographql/skills, covering preamble, primary directive, voice, language and framing.
skills apollo-client-skill.instructions.md
Instructions for apollographql/skills, covering apollo client skill writing instructions, general rules, rules for discussing fragments, rules for code examples and external resources.
skills javascript-formatting.instructions.md
Instructions for apollographql/skills, covering javascript/typescript code formatting instructions, code formatting, formatting before commits, code blocks in markdown and skills covered.
skills AGENTS.md
Instructions for apollographql/skills, covering claude.md, repository overview, repository structure, skill file format and validation.
skills CLAUDE.md
Instructions for apollographql/skills, a project described as: Apollo GraphQL Agent Skills.
payload CLAUDE.md
Instructions for payloadcms/payload, covering claude.md, project structure, key directories, architecture notes and quick start.