Summary
After an epoch unlock, the API can detect the Timeline screen and expose advance / back, but menu_select with option: advance stops at manual_action_required: true instead of completing the reveal. This leaves agents unable to continue from the main menu without a manual UI click.
Current observed state
From the active game session:
{
"state_type": "menu",
"menu_screen": "timeline",
"options": [
{ "name": "advance", "enabled": true },
{ "name": "back", "enabled": true }
]
}
Calling:
{ "action": "menu_select", "option": "advance" }
returns:
{
"status": "ok",
"message": "Epoch unlocks are obtained but not revealed; not forcing timeline reveal from automation",
"pending_epoch_ids": ["IRONCLAD5_EPOCH"],
"manual_action_required": true,
"done": true
}
Why this matters
The previous Timeline work correctly avoided an unsafe path that could produce invalid unlock-state errors, but the resulting behavior still blocks long-running agents after unlocks. Agents can recognize that Timeline is active, but cannot progress the required reveal flow through the API.
Desired behavior
The API should provide a safe, explicit way to advance pending Timeline epoch reveals without requiring mouse input. The implementation should avoid the known invalid unlock-state path, but should still let an agent continue once the game is already on the Timeline reveal flow.
Acceptance criteria
menu_select timeline continues to report manual_action_required if opening Timeline from the main menu would be unsafe.
- Once the game is already on the Timeline screen,
menu_select advance should safely handle a pending obtained epoch reveal where possible.
- The response should clearly distinguish retryable animation/busy states, completed reveals, and states that still require manual input.
- Documentation should describe the safe Timeline reveal behavior and its remaining limits.
- Add targeted coverage for the Timeline guard logic if practical in this codebase.
Summary
After an epoch unlock, the API can detect the Timeline screen and expose
advance/back, butmenu_selectwithoption: advancestops atmanual_action_required: trueinstead of completing the reveal. This leaves agents unable to continue from the main menu without a manual UI click.Current observed state
From the active game session:
{ "state_type": "menu", "menu_screen": "timeline", "options": [ { "name": "advance", "enabled": true }, { "name": "back", "enabled": true } ] }Calling:
{ "action": "menu_select", "option": "advance" }returns:
{ "status": "ok", "message": "Epoch unlocks are obtained but not revealed; not forcing timeline reveal from automation", "pending_epoch_ids": ["IRONCLAD5_EPOCH"], "manual_action_required": true, "done": true }Why this matters
The previous Timeline work correctly avoided an unsafe path that could produce invalid unlock-state errors, but the resulting behavior still blocks long-running agents after unlocks. Agents can recognize that Timeline is active, but cannot progress the required reveal flow through the API.
Desired behavior
The API should provide a safe, explicit way to advance pending Timeline epoch reveals without requiring mouse input. The implementation should avoid the known invalid unlock-state path, but should still let an agent continue once the game is already on the Timeline reveal flow.
Acceptance criteria
menu_select timelinecontinues to reportmanual_action_requiredif opening Timeline from the main menu would be unsafe.menu_select advanceshould safely handle a pending obtained epoch reveal where possible.