Bagman

Secure key management for AI agents. Use when handling private keys, API secrets, wallet credentials, or when building systems that need agent-controlled fun...

This skill appears to be what it says: a set of patterns and examples for using 1Password to manage agent secrets. Before installing/use: (1) verify you have the official 'op' CLI from 1Password and understand how to sign in (the skill assumes an active session/OP_SESSION token), (2) confirm that any '.env' or env-file referenced is a template that does not contain real secrets, (3) limit agent session keys to minimal permissions and short lifetimes as recommended, (4) audit 1Password access logs and restrict the agent's vault/item permissions, (5) test the provided sanitize_output routines for false positives/negatives in your environment, and (6) be cautious about giving an autonomous agent any ability to request new session keys without human approval. If you need higher assurance, ask the author to document the authentication prerequisites (how to safely obtain OP_SESSION tokens) and to clarify the env-file example so it cannot be misused to write secrets to disk.

Type: OpenClaw Skill Name: master-skills Version: 0.1.0 The 'bagman' skill bundle is a security-focused resource providing patterns and code snippets for secure secret management in AI agents. It includes defensive measures such as output sanitization (regex-based redaction of keys), prompt injection detection, and the use of 1Password CLI (op) for runtime secret retrieval. The content in SKILL.md is educational and aimed at preventing accidental data leaks and unauthorized access, with no evidence of malicious intent or harmful instructions.