Model Autopilot
IDE-aware step-by-step model switching with a mandatory STOP banner before every switch. Reads .project-ai/MODEL_AUTOPILOT.yaml. Resolves the active IDE (cursor, vscode, claude, windsurf, jetbrains, neovim, xcode, cli) and uses that IDE's actual picker labels (Opus 4.7, GPT-5 Codex, Cascade Base, Composer 2.5, claude-opus-4-7, …) from @enterprise-skills/core IDE_MODEL_MAPS. Forces a Model check verify before every step so the agent can't silently advance when the user is on the wrong picker.
Expected outcome
IDE-aware step-by-step model switching with a mandatory STOP banner before every switch. Reads .project-ai/MODEL_AUTOPILOT.yaml. Resolves the active IDE (cursor, vscode, claude, windsurf, jetbrains, neovim, xcode, cli) and uses that IDE's actual picker labels (Opus 4.7, GPT-5 Codex, Cascade Base, Composer 2.5, claude-opus-4-7, …) from @enterprise-skills/core IDE_MODEL_MAPS. Forces a Model check verify before every step so the agent can't silently advance when the user is on the wrong picker.
Typical deliverable
A policy decision, enforcement result, and traceable governance record.
01
Inspect context
Loads project authority and checks prerequisites.
02
Run the procedure
Executes defined steps and honors safety gates.
03
Return evidence
Produces a reviewable result instead of an unsupported claim.
Licensed implementation
The licensed module adds trigger phrases, CLI aliases, dependency and coordination maps, and its step-by-step governed procedure.
No per-skill purchases: a plan unlocks every skill in its tier, this one included.