Support

Diagnostics and escalation

Start every support request with a request ID, org slug, repo ID, agent name, daemon status, and the command or URL that failed. Do not send raw API keys or Clerk secrets.

Diagnostics

Production health

curl -sS https://collective-self.vercel.app/health

Local daemon

curl -sS http://127.0.0.1:8787/status

Context query

cargo run -p collective-cli -- ask "Fix enterprise users being logged out randomly"

Verification

pnpm lint && pnpm typecheck && pnpm test && go test ./... && pnpm build

Support packet

Request ID from the failed response.
Org slug and repo ID.
Agent integration mode: CLI, MCP, local HTTP, or dashboard.
Recent audit action and timestamp.
Sanitized logs with secrets removed.

Service boundaries

Local mode uses the Rust daemon and SQLite cache. Production mode uses Vercel Functions for the Go API and Clerk for launch auth. Neo4j, Qdrant, ClickHouse, NATS, and Temporal remain documented fallback integrations until live credentials are configured.