.sheevactions JSON using keyed source objects, nested source maps, flat placeholders, and an explicit execute.type. Packaged libraries can place the completed manifest at the root of a .sheevactionszip file.
How does your app communicate?
Choose how SheevChat will talk to the app you want to control.
How does SheevChat find the app?
Choose the runtime's connection or discovery pattern. All file paths remain inside approved local roots.
Do your actions need dynamic dropdowns?
Dropdowns let users pick options fetched live from the app — like a list of scenes, models, or hotkeys.
What WebSocket message style does your app use?
This only applies to WebSocket connections. Different apps format their messages differently — this affects how SheevChat maps responses.
Does your app require authentication?
Some local apps require a token or session handshake before dropdowns and actions will work.
statusRequest and statusMap before starting a new authentication flow.
persist: false. The token will only be used during the current authentication flow.
What should the library be called?
These values become the manifest identity shown in SheevChat. You can change them later.
Required fields must be completed before continuing.
Where does SheevChat connect?
Provide the connection or safe discovery details selected in Step 2.
Required for the selected connection pattern.
...What should the starter action do?
Name the first action and provide its request target. Dropdown list endpoints and response mappings remain marked TODO because they depend on the target API.
Required fields define the starter action.
Guided fill-ins
Unresolved template values
Sources and response lists
Actions, payloads, and results
Advanced JSON editing
Live manifest preview
Drafts may be downloaded with unresolved values for AI-assisted completion. A manifest is marked ready only when required TODO values and schema errors are gone. See the AI-friendly Action Library Wiki for the full schema.