← ClaudeAtlas

skills-keyslisted

Manage API keys for the runner's --execute layer. CRUD on ~/.skills.env (chmod 600): list / add / update / remove / verify / export. Covers OpenAI, Gemini, BFL, FAL, Replicate, Runway, Kling, Suno, Eleven, Ideogram, Anthropic, S3. Use when: 'add my OpenAI key', 'rotate the Suno key', 'which keys are set', 'verify my key works'.
Mikefluff/skills · ★ 19 · AI & Automation · score 79
Install: claude install-skill Mikefluff/skills
<objective> Provide a thin CLI on top of `~/.skills.env` (the runner's key store) so users can manage keys WITHOUT editing dotfiles by hand. Values are stored plaintext but chmod 600. Each runner reads the file at startup, merging entries into `os.environ` without overriding explicit shell exports. This skill does NOT: - Store keys in macOS Keychain / Linux Secret Service (cross-platform parity matters more than OS-native vault for v1). - Encrypt the file at rest (chmod 600 is the v1 posture). - Manage non-runner keys (database creds, SSH keys, etc.). - Validate every provider — only the 9 that have lightweight public verify endpoints. Use when the user wants to add a new API key, swap one, see what's set, or check whether a key is actually valid against the vendor. </objective> ## ROLE Read the user's intent (add/update/remove/list/verify/enable/disable/path/export) → call the matching `common/runners/cli/keys.py` subcommand → relay the result. Mask values when displaying. Prompt for value silently if not provided on the command line. Never write keys to stdout in full unless `export` is explicitly requested. ## PIPELINE 1. **Parse intent**: which subcommand and which arg(s). 2. **Run** `python3 scripts/run.py <subcommand> [args]`. 3. **Relay the output** — already masked / colorized by the CLI. 4. **Optional follow-up**: if user added a new key, suggest running `skills-keys verify <KEY>` to confirm it works. ## MODES ### Inspection - `skills-keys list` — show all k