Verified Agent Identity

Billions/Iden3 authentication and identity management tools for agents. Link, proof, sign, and verify.

What to consider before installing: - This package creates and stores private keys unencrypted in $HOME/.openclaw/billions/kms.json (and other identity files). If you install, expect long-lived plaintext key material on disk. Consider running in a sandbox, using an encrypted KMS, or adjusting file permissions (and verify permissions are actually enforced). - The skill will call external services (Billions RPC, attestation-relay, a DID resolver). The protocol embeds signed tokens into callback URLs; review those endpoints and privacy implications before using with production keys or real assets. - The repository expects Node >=20 and the openclaw CLI on PATH. The registry metadata omitted those required binaries — ensure you have the correct runtime installed and trust the openclaw binary before allowing it to send messages. - Installation requires running `npm install` (package-lock is provided). That pulls standard npm packages; consider auditing dependencies if you require high assurance. - The code validates inputs and uses execFileSync to call a fixed `openclaw message send` command; it applies tokenization and regex filtering to reduce shell-injection risk — but validation and sanitization are area to review if you expect untrusted input. - If you are not comfortable with plaintext key storage or with signing callbacks being sent to third-party relays, do not install or only use test identities in a controlled environment. If you want, I can list the exact files that write or read key material and suggest specific hardening changes (e.g., encrypt kms.json, set restrictive file modes, or switch to an OS-provided secure key store).

Type: OpenClaw Skill Name: verified-agent-identity-5 Version: 0.1.0 The skill bundle provides a decentralized identity (DID) management toolkit for AI agents on the Billions Network using the Iden3 protocol. It facilitates creating identities, signing challenges, and linking agent DIDs to human owners. While it stores private keys unencrypted in '$HOME/.openclaw/billions/kms.json' (a significant security vulnerability), this behavior is explicitly documented in the README and SKILL.md as a known design choice. The implementation includes robust security controls, such as strict input sanitization using 'shell-quote' and regex validation in 'scripts/shared/utils.js' to prevent shell injection when calling the 'openclaw' CLI. No evidence of data exfiltration, malicious persistence, or deceptive prompt injection was found.