embedding-ft

Installation
SKILL.md

Embedding FT

Run Workflow

  1. The wrapper uses the repository-pinned published CLI. Use TIANGONG_LCA_CLI_DIR or --cli-dir only for an exact local dev/CI override.
  2. Configure the API base URL, Supabase publishable key, and environment-specific public OAuth client ID.
  3. Run tiangong-lca auth doctor-auth --json. If it returns login-required, stop and ask the human user to run tiangong-lca auth login in a trusted terminal.
  4. Execute node scripts/run-embedding-ft.mjs with standard tiangong-lca admin embedding-run flags.
  5. The wrapper delegates to tiangong-lca admin embedding-run; the CLI resolves and refreshes its private OAuth session.
  6. Inspect completedJobs and failedJobs, then triage via references.

Never ask an AI user for a username, password, authorization code, access token, refresh token, or the deprecated encoded API key. Headless runs may use only an orchestrator-injected short-lived TIANGONG_LCA_ACCESS_TOKEN; do not pass it through argv, prompts, logs, or artifacts.

Commands

TIANGONG_LCA_API_BASE_URL="https://example.supabase.co/functions/v1" \
TIANGONG_LCA_SUPABASE_PUBLISHABLE_KEY="<publishable-key>" \
TIANGONG_LCA_OAUTH_CLIENT_ID="<public-oauth-client-id>" \
tiangong-lca auth doctor-auth --json
Installs
17
GitHub Stars
6
First Seen
Feb 2, 2026
embedding-ft — tiangong-lca/skills