@sansaai/setup
v0.3.0
Published
Configure coding agents to use Sansa as the LLM provider with model sansa-auto.
Maintainers
Readme
@sansaai/setup
Configure coding agents to use Sansa as the LLM provider with model
sansa-auto. The command stores the API key under ~/.sansa, writes
OpenAI-compatible provider settings for Codex and OpenCode, and installs a
sansa skill for Claude Code, Cursor, Codex, and OpenCode.
Usage
SANSA_API_KEY=sk-... npx -y @sansaai/setup
npx -y @sansaai/setup --api-key sk-...
npx -y @sansaai/setup -y --api-key sk-...
npx -y @sansaai/setup --api-key sk-... -a claude-code
npx -y @sansaai/setup --restoreThe key is read from --api-key or SANSA_API_KEY. Setup does not prompt for
it. Interactive runs (a TTY, without -y) ask which agents should use Sansa as
the model provider and which should get the skill. -y skips every prompt,
which is the path a coding agent uses when it runs this command from an
onboarding prompt.
Without an explicit -a/--agent, -y only writes credentials and installs
the skill for detected agents — it never rewrites Codex or OpenCode provider
settings on its own. Pointing Codex or OpenCode at Sansa requires either
naming them explicitly (-a codex, -a opencode) or an interactive run, so
an unattended agent can never flip where those tools send every request
without you asking for it.
What it writes
User-level credentials (mode 0600):
~/.sansa/credentials(SANSA_API_KEY=...)~/.sansa/api-key(raw key, read by Codex and OpenCode)
Provider config, with sansa-auto as the default model:
- Codex:
~/.codex/config.toml - OpenCode:
~/.config/opencode/opencode.json
The bundled sansa skill is copied into the selected agents' skill directories.
--restore reverts Codex and OpenCode provider files from the first backup
under ~/.sansa/backups. Credentials, project env writes, and skills are left
in place.
