Scan the machine
Detect installed CLIs, MCP servers, agent skills, and provider routes.
The Brain detects local CLIs, agent skills, MCP servers, and provider routes Genlock can use — without ever taking ownership of third-party credentials.
For creators who want AI power that is observable, not magical.
/brainA fused verdict banner — Ready to work / N blockers / Blocked — with BONE readiness counts.
/brainRun Preflight probes runtimes + providers and writes capability_matrix.json with honest per-capability badges.
/integrationsConnection health, setup checklist, and provider defaults.
/ai-inspectorSkill coverage, provider routes, and the workflow map.
/codexFirst-class OpenAI Codex CLI cockpit (when detected).
/claudeFirst-class Claude Code CLI cockpit (when detected).
/higgsfieldThe native generation cockpit.
Open dashboard →/antigravityDetected agy CLI command surface (when detected).
/openartOpenArt MCP read/schema cockpit via Codex.
Detect installed CLIs, MCP servers, agent skills, and provider routes.
Group tools into What Works Now, What Needs Attention, and Agent Routes.
Surface install and auth steps without storing any tokens.
Jump straight into Codex, Claude, Higgsfield, OpenArt, or Antigravity.
Codex, Claude, Antigravity, Higgsfield, and OpenArt MCP show up as live capabilities, not static claims.
One Run Preflight probes composition runtimes and providers, then answers "can I ship today?" with a Healthy / Degraded / Blocked health strip and structured quick-setup steps.
Each BONE's requires[] joins against the matrix — armed / needs-setup / unavailable — and every blocker gets two paths: Fix this now, or Tell the LLM with a grounded handoff prompt.
Auth stays in Codex, Claude, provider CLIs, or MCP OAuth — Genlock Studio never becomes a credential store.