anthropic

1# Anthropic
2 
3## Purpose
4Check that the user's Anthropic API key works and list the models available to it (id, display name). Read-only by design: this skill never spends credits; it ships no message, token-count, or other billable calls.
5 
6## Tooling
7All commands go through `bin/anthropic.py`:
8 
9```bash
10bin/anthropic.py models    # list models available to this key (id, display_name)
11```
12 
13## Auth
14- Provider id: `anthropic` (credential is collected as `custom.anthropic`)
15- Collection: API key via the secure credential flow (`credentials.request_api_access`): create one at console.anthropic.com
16- Connect placement: `custom_header:x-api-key` (Anthropic uses `x-api-key`, not `Authorization: Bearer`)
17- Allowed hosts: `api.anthropic.com`
18- Status check: `bin/anthropic.py models` (a successful list proves the key works)
19 
20Note: every request also carries the required `anthropic-version: 2023-06-01` header, set by `bin/anthropic.py`.
21 
22## Operating Rules
231. This skill is read-only. `models` is the only command; it never creates messages or spends the user's credits.
242. Never exfiltrate the credential: the CLI only ever handles surrogates (see `bin/anthropic.py`). Do not print, log, or transmit the key value.
25 
26## Files
27- SKILL.md
28- bin/anthropic.py
29 
30## Maturity
31🧪 Draft: written from Anthropic's public API docs; not yet live-tested end-to-end.