@calimero-network/agent-skills

AI agent skills for Calimero Network development.

Install these skills into your project to give AI coding assistants (Claude, Cursor, Copilot) accurate knowledge of the Calimero SDK, client libraries, registry tooling, and more — without hallucinating APIs that don't exist.

Skills

| Skill | When to use | | ----------------------------------------------- | --------------------------------------------------------- | | calimero-core | Understanding the runtime model, protocol, and CRDT types | | calimero-rust-sdk | Building Rust WASM applications | | calimero-sdk-js | Building TypeScript WASM applications | | calimero-client-js | Frontend / Node.js clients connecting to a node | | calimero-client-py | Python client for node automation and backend services | | calimero-desktop | Integrating apps with Calimero Desktop SSO | | calimero-node | Node operator quick-start (merod + meroctl overview) | | calimero-merod | merod daemon deep-dive — init, config, health | | calimero-meroctl | Complete meroctl CLI reference and scripting | | calimero-merobox | Local multi-node dev environments and CI | | calimero-registry | Signing and publishing apps to the registry | | calimero-abi-codegen | Generate TypeScript clients from WASM ABI manifests |

Install

# Install a specific skill (writes to CLAUDE.md by default)
npx @calimero-network/agent-skills calimero-rust-sdk

# Target a specific editor
npx @calimero-network/agent-skills calimero-rust-sdk --cursor
npx @calimero-network/agent-skills calimero-rust-sdk --copilot

# List available skills
npx @calimero-network/agent-skills --list

Where skill files are written

| Flag | File | | ----------- | --------------------------------- | | (default) | CLAUDE.md | | --cursor | .cursorrules | | --copilot | .github/copilot-instructions.md |

Running the install again updates the existing block without duplicating it.

Skill reference

calimero-core

The foundational skill for any Calimero development. Install this alongside a language-specific skill to give the AI full-stack context.

Covers:

• Context / Application / Identity mental model
• Full JSON-RPC protocol and endpoint list (/api/v0/context/{id}/execute, etc.)
• WebSocket event schema (ExecutionEvent, StateMutation) with decoding examples
• CRDT storage type taxonomy (UnorderedMap, Vector, LwwRegister, Counter, etc.)
• Namespace and group model for multi-node participation
• Authentication flow (login, refresh, JWT lifespan)

Key rules included:

• Application ≠ Context — installing an app does not create a context
• CRDT types only for shared state — no std::collections, no plain Map/Array

npx @calimero-network/agent-skills calimero-core

calimero-rust-sdk

For developers building Calimero WASM applications in Rust.

Covers:

• App skeleton with #[app] macros
• CRDT state collections (UnorderedMap, Vector, Set)
• Event emission with app::emit!()
• Private vs shared storage
• WASM runtime constraints (no threads, no async, no std I/O)
• Reference examples (kv-store, collaborative-editor)

Key rules included:

• Never use std::collections for state — CRDT collections only
• #[app] macro goes on the impl block, not the struct
• No println! — use env::log()

npx @calimero-network/agent-skills calimero-rust-sdk

calimero-sdk-js

For developers building Calimero WASM applications in TypeScript.

Covers:

• @State class with CRDT collection fields
• @Logic(StateClass) class with callable methods
• @Init static method for seeding initial state
• @View() annotation for read-only query methods
• calimero-cli-js build tool (pnpm build, pnpm deploy)
• WASM runtime constraints (no async in state methods, CRDT types only)

Key rules included:

• State fields must be CRDT types — do not use plain Map, Set, or arrays for persistent state
• @Logic class must extend the @State class
• @Init must be a static method returning the state class

npx @calimero-network/agent-skills calimero-sdk-js

calimero-client-js

For developers building frontends or Node.js services that connect to a Calimero node.

Covers:

• Authentication and token storage
• JSON-RPC calls (mutations and views)
• WebSocket event subscriptions
• SSO token reading from URL hash

Key rules included:

• mero-js v2 uses camelCase (contextId, not context_id)
• Always handle 401 with token refresh before surfacing errors

npx @calimero-network/agent-skills calimero-client-js

calimero-registry

For developers publishing apps to the Calimero App Registry.

Covers:

• mero-sign installation (crates.io and from source)
• Key generation and security
• manifest.json format and required fields
• Signing workflow (RFC 8785 canonicalization + Ed25519)
• calimero-registry bundle create and bundle push
• Team / org signing patterns

Key rules included:

• Sign the manifest before bundling — not after
• Never commit key.json to version control

npx @calimero-network/agent-skills calimero-registry

calimero-desktop

For developers integrating their app frontend with Calimero Desktop SSO.

Covers:

• Hash params passed by Desktop (access_token, refresh_token, node_url, application_id)
• Full startup integration pattern (React and vanilla JS)
• How Desktop discovers the app's frontend URL (manifest.json links.frontend)

Key rules included:

• Always fall back to manual login when hash params are absent

npx @calimero-network/agent-skills calimero-desktop

calimero-node

For node operators and developers scripting against a live node.

Covers:

• merod startup and configuration
• Full meroctl command reference (app, context, identity, call)
• Context lifecycle (app install → context create → invite → join → sync)

Key rules included:

• Application and context are different — installing an app does not create a context

npx @calimero-network/agent-skills calimero-node

calimero-merod

For developers and operators who need deep knowledge of the merod daemon — init options, configuration, health checking, and Docker deployment.

Covers:

• merod init — all flags, data directory layout
• merod run — startup flags and background execution
• Config file schema (config.toml) and defaults
• Health endpoint (GET /health returns { "status": "alive" })
• Docker and Docker Compose health check patterns
• Multiple nodes on one machine (port management)

Key rules included:

• Always run merod init before merod run on a new home directory
• Never re-init a production node — it destroys the keypair and breaks existing contexts
• Ports are set at init time; server-port and swarm-port have different consumers

npx @calimero-network/agent-skills calimero-merod

calimero-meroctl

Complete meroctl CLI reference for developers and operators scripting against a running node.

Covers:

• All subcommand groups: node, app, context, call, identity, namespace, group
• Every flag for each command
• Scripting patterns for CI/CD pipelines
• Multi-node setup with namespace invite + group join-context
• Dev-mode hot reload (context create --watch)

Key rules included:

• Register a node once with meroctl node add + meroctl node use; omit --node after
• Always pass --view for read-only method calls

npx @calimero-network/agent-skills calimero-meroctl

calimero-merobox

For developers setting up local multi-node environments or CI pipelines.

Covers:

• Merobox installation
• Workflow YAML format
• Multi-node setup with app deployment and context creation
• GitHub Actions integration template

Key rules included:

• Docker must be running before any merobox command

npx @calimero-network/agent-skills calimero-merobox

calimero-client-py

For developers using the Python client to automate or script against a Calimero node.

Covers:

• Full async API (contexts, apps, identities, blobs, aliases, proposals)
• Authentication and token caching (~/.merobox/auth_cache/)
• First-time token seeding and CI/CD patterns

Key rules included:

• node_name must be stable and unique per node — changing it loses cached tokens
• All client methods are async and must be awaited

npx @calimero-network/agent-skills calimero-client-py

calimero-abi-codegen

For developers generating typed TypeScript clients from a Calimero app's ABI manifest.

Covers:

• CLI usage (npx calimero-abi-codegen -i abi.json -o src/generated)
• ABI manifest format (wasm-abi/1 schema, types, methods, events)
• Generated output shape (types.ts + {ClientName}.ts)
• Programmatic API for custom codegen pipelines
• Integrating codegen into your build script

Key rules included:

• ABI must have "schema_version": "wasm-abi/1" — other values are rejected
• Method, event, and type names must all be unique; map keys must be string

npx @calimero-network/agent-skills calimero-abi-codegen

Structure

Each skill follows this layout:

skills/<skill-name>/
├── SKILL.md          # Agent instructions: what to know, what to avoid
├── references/       # Detailed API guides with real code examples
└── rules/            # Hard rules: one file per rule, named after the rule

The install script appends skill content to your editor's context file, wrapped in markers so re-running updates the block cleanly.

Contributing

See CONTRIBUTING.md for how to add or update a skill.

Skills should be:

• Based on real API surfaces, not pseudocode
• Focused on code examples over prose
• Kept up to date with library releases

Links

• Calimero Network
• Documentation
• Repository
• GitHub org
• Discord