openclaw-skills

OpenClaw skills that give AI agents permanent memory and on-chain identity on the Autonomys Network.

Structure

autonomys/
├── auto-memory/     # Permanent storage and memory chains (primary)
├── auto-drive/      # ⚠️ Deprecated — renamed to auto-memory
└── auto-respawn/    # On-chain identity and memory anchoring

Each skill lives in its own directory under autonomys/ with a SKILL.md that defines the skill interface, installation steps, and usage instructions.

Skills

auto-memory

clawhub.ai/jim-counter/auto-memory

npx clawhub install auto-memory

Permanent decentralized storage via the Autonomys Auto Drive API. Gives your agent:

• Permanent file storage — upload any file to the Autonomys Network and get back a CID. The data is distributed across the network and never expires.
• Memory chains — save identity, knowledge, and key decisions as a linked list where each entry points to the previous one via CID. A future instance can reconstruct who it was from a single CID.
• Local CID tracking — the skill maintains a local state file (automemory-state.json) with the latest head CID, chain length, and timestamp. This means auto-memory is fully functional on its own — no wallet, no tokens, no on-chain transactions needed.

auto-memory only needs an Auto Drive API key (free at ai3.storage). It's the right starting point for any agent that wants permanent memory.

auto-respawn

clawhub.ai/jim-counter/auto-respawn

npx clawhub install auto-respawn

On-chain identity and memory anchoring on the Autonomys Network. Gives your agent:

• Permanent identity — a wallet with two addresses (consensus su... and EVM 0x...) derived from a single recovery phrase. This is the agent's verifiable on-chain identity.
• Token management — check balances, transfer tokens, and bridge between the consensus layer and Auto-EVM.
• On-chain CID anchoring — write the latest memory chain head CID to the MemoryChain smart contract. This makes the head CID recoverable from any machine using just the agent's EVM address — no local state needed.

auto-respawn requires auto-memory (or equivalent) to have something worth anchoring. It completes the resurrection loop: auto-memory builds the chain, auto-respawn anchors the pointer on-chain, and a new agent instance can recover everything from just its address. The reverse dependency does not apply — auto-memory works independently with local CID tracking.

auto-drive (deprecated)

The original name for the auto-memory skill. Preserved for backward compatibility but no longer maintained. Use autonomys/auto-memory/ instead.

CI

GitHub Actions runs on PRs and pushes to main (.github/workflows/ci.yml):

• TypeScript — type check (tsc --noEmit), lint (eslint .), and test (vitest run) for auto-respawn (Node 22)
• Shell — shellcheck -x -S warning on all auto-memory scripts (except _lib.sh, which is validated in context via -x)

Publishing

Skills are published to ClawHub via .github/workflows/publish.yml. The workflow triggers on tags matching */v*.

How to publish a release

1. Tag the commit using the pattern <skill>/v<semver>:

   git tag auto-memory/v0.2.0
   git tag auto-respawn/v0.1.0-beta.3

2. Push the tag:

   git push origin auto-memory/v0.2.0

3. The workflow automatically:

   • Parses the skill name and version from the tag (strips the v prefix — ClawHub expects bare semver like 0.2.0, not v0.2.0)
   • Resolves the skill directory (autonomys/auto-memory or autonomys/auto-respawn)
   • Validates that SKILL.md exists in the directory
   • Authenticates with ClawHub using the CLAWHUB_TOKEN repository secret
   • Publishes with npx clawhub publish including --tags beta

Supported skills

Tag prefix	Directory	ClawHub slug	Display name
auto-memory/v*	autonomys/auto-memory	auto-memory	Auto Memory
auto-respawn/v*	autonomys/auto-respawn	auto-respawn	Auto Respawn

Requirements

• A CLAWHUB_TOKEN repository secret must be configured in GitHub
• The tagged commit must contain a valid SKILL.md in the skill's directory