aa-sdk
AGENTS.md321 starsChanged 5 months agoMITView on GitHub
AI assistants should load skills from .agents/skills/ based on the task context. The v5 SDK now lives on main (graduated from v5.x.x on 2026-05-20). Use main as the audit source, comparison base, and PR base unless the user explicitly asks for another branch. The legacy v4 SDK lives on the v4.x.x branch and only receives backports/security fixes. Do not treat untracked top-level directories as source truth. On main, the tracked workspace is the root config plus .agents/, .claude/, .cursor/, .github/,…
- Installs packages
We split this AGENTS.md into the rules it gives an agent
2 rules. About 0 are common, 0 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 9
Do not treat untracked top-level directories as source truth.
- Very specific: line 101
Check docs/solutions/ for gotchas before changing workflows, generated docs, tests, or package boundaries.
≈ “Common” is an estimate: at least 1 in 5 of the 36,368 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Agent Router
AI assistants should load skills from .agents/skills/ based on the task
context. The v5 SDK now lives on main (graduated from v5.x.x on
2026-05-20). Use main as the audit source, comparison base, and PR base
unless the user explicitly asks for another branch. The legacy v4 SDK lives
on the v4.x.x branch and only receives backports/security fixes.
Do not treat untracked top-level directories as source truth. On main, the
tracked workspace is the root config plus .agents/, .claude/, .cursor/,
.github/, .vitest/, docs/, halp/, packages/, scripts/, and
templates/.
Quick Reference
pnpm install
pnpm run build:libs
pnpm run lint:check
pnpm run lint:write
pnpm run test:ci
pnpm run test:typecheck
./scripts/run-affected-tests.sh main
pnpm run docs:sdk
pnpm run lint:docs
pnpm run docs:broken-linksFor Alchemy engineers, run npx turbo login --sso-team=alchemy-dot-com once to
use remote caching.
Available Skills
| Skill | Description |
tooling | pnpm, Turborepo, Lerna, local validation, and generated-output boundaries. |
package-exports | Package export maps, ESM import paths, public/internal/experimental surfaces. |
error-handling | SDK error hierarchy, BaseError, and package error conventions. |
wallet-apis | Wallet API client, Zod codecs, signer/client model, and Solana support. |
smart-accounts | Light Account, Modular Account v1/v2, modules, permissions, and deferred actions. |
testing | Vitest projects, Anvil/Foundry/Rundler setup, and affected test execution. |
documentation | MDX docs, TypeDoc reference generation, docs/docs.yml, and code snippet preservation. |
release-workflows | npm publish (publish-package.yml) and post-merge docs hooks (revalidate, indexer). |
Skill Loading
| Task | Skills to load |
Adding or changing package exports, barrels, or package.json exports | package-exports, tooling |
Editing TypeScript source in packages/ | package-exports, error-handling, plus package-specific skills |
| Working on Wallet API client, transport, actions, schemas, or codecs | wallet-apis, error-handling, testing |
| Working on account implementations, modules, permissions, or deferred actions | smart-accounts, error-handling, testing |
| Writing or modifying tests | testing, tooling |
Editing .vitest/, vitest.config.ts, or test setup | testing, tooling |
| Editing docs or TypeDoc comments | documentation |
Editing docs/pages/reference/, typedoc.json, or docs/docs.yml | documentation, tooling |
Editing .github/workflows/, lerna.json, version injection, or publish automation | release-workflows, tooling |
| Creating a PR or evaluating deploy/release safety | release-workflows, tooling |
Directory Mapping
| Path | Relevant skills |
packages/common/ | package-exports, error-handling, testing |
packages/aa-infra/ | package-exports, error-handling, testing |
packages/smart-accounts/ | smart-accounts, package-exports, error-handling, testing |
packages/wallet-apis/ | wallet-apis, package-exports, error-handling, testing |
.vitest/, vitest.config.ts, packages/*/vitest.config.ts | testing, tooling |
templates/, halp/, turbo.json, pnpm-workspace.yaml, package.json | tooling |
docs/**/*.mdx, docs/docs.yml, typedoc.json, tsconfig.typedoc.json | documentation |
docs/pages/reference/ | documentation |
.github/workflows/, .github/actions/, lerna.json, packages/*/inject-version.ts | release-workflows, tooling |
Project Structure
aa-sdk/
├── .agents/ # Agent skills, commands, and repo guidance
├── packages/
│ ├── common/ # Shared transport, chain, logging, config, and errors
│ ├── aa-infra/ # Rundler fee estimation and RPC types
│ ├── smart-accounts/ # Light Account, Modular Account, modules, permissions
│ └── wallet-apis/ # High-level Wallet API viem client and actions
├── .vitest/ # Shared Vitest setup, Anvil/Rundler fixtures, test helpers
├── docs/ # Fern-style docs and TypeDoc-generated reference MDX
├── scripts/ # TypeDoc navigation, docs sync, affected-test helpers
├── templates/ # Shared TypeScript and ESLint workspace packages
├── halp/ # Internal package scaffolding CLI
└── .github/ # PR checks, publish workflows, docs sync, and instructionsSlash Commands
| Command | Description |
/docs-reviewer | Existing Claude command for MDX documentation review. |
/publish-package | Checklist for dry-running or publishing via the publish-package.yml flow. |
Documentation Discovery
- Check
docs/solutions/for gotchas before changing workflows, generated docs, tests, or package boundaries. - Trust source files, package configs, scripts, and workflow YAML over stale prose.
- Existing docs AI guidance lives in
.claude/commands/docs-reviewer.md,.cursor/rules/docs-reviewer.mdc,.github/copilot-instructions.md, and.github/instructions/docsreview.instructions.md.
Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.
Of the AGENTS.md files we read, the 28,693 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repository16 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- .github/copilot-instructions.mdCopilot instructions
- .cursor/rules/docs-reference.mdcCursor rule
- .cursor/rules/docs-reviewer.mdcCursor rule
- .cursor/rules/error-handling.mdcCursor rule
- .cursor/rules/package-exports.mdcCursor rule
- .cursor/rules/release-workflows.mdcCursor rule
- .cursor/rules/testing.mdcCursor rule
- documentationSkill
- error-handlingSkill
- package-exportsSkill
- release-workflowsSkill
- smart-accountsSkill
- testingSkill
- toolingSkill
- wallet-apisSkill
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.