Bitwarden

**Quick judgment**: Useful, powerful, and inherently high-risk. This skill makes your Bitwarden/Vaultwarden vault accessible to agents via the `rbw` CLI. It is suitable only for tightly controlled environments where you fully trust the agent workflows and understand how outputs are stored and used. **What it does well** - Provides solid coverage of everyday vault operations: list, search, get (including full JSON), add, and sync. - Uses the well-established `rbw` CLI rather than ad‑hoc API calls, which is good for correctness and security posture. - Works with both Bitwarden cloud and Vaultwarden, with configurable base URL. **Key risks / cautions** - **High impact of leakage**: Any credentials retrieved via this skill can be exposed to other tools, logs, or prompts if workflows are not carefully designed. - **Exec-based command execution**: It runs `rbw` via `exec`; misconfiguration or overly broad tool access could allow commands or outputs you did not anticipate. - **Interactive unlock complexity**: Unlocking via `pinentry-curses` and `tmux` can be fragile; failures or misconfiguration can break flows or encourage insecure workarounds. - **Prompt / tool misuse**: Ambiguous prompts could cause the agent to retrieve more secrets than intended (e.g., listing large portions of the vault instead of just one item). **Recommended scenarios** - You’re on a **single-user, secure machine**, already using `rbw`, and want automated but controlled access to specific credentials for scripts or dev tooling. - You operate a **self-hosted Vaultwarden** and prefer a local CLI-based integration rather than cloud API usage. - You can enforce **narrow, audited workflows** (e.g., only retrieving a specific item and passing it directly into a known tool without logging or re-printing it). Avoid using this skill in shared, poorly isolated, or heavily logged environments, and do not grant it to general-purpose agents that aren’t tightly constrained in how they handle and store secret outputs.