Expand description
The methods of the contract, all called by the core.
Constants§
- ANSWER_
PERMISSION - The answer to a permission or a question the harness asked (
AnswerPermission,AnswerQuestion→Writes). - ANSWER_
QUESTION - ATTACH
- An agent given to the plugin, before its launch and again after a restart (
Attach). - CLOSE
- The agent stops using its shared harness (
Close, a call or a notification): gently, its turn finished first andItem::Closedsaid once it is; or at once, its turn stopped. - COMPACT
- Compact its context (
Compact→Compacted), asEngine::compactionsays it can. - CONTROL
- A request in the harness’s own protocol, its answer: for the tests, and what has no method yet (
Control). - DETACH
- The agent’s process is gone: its state may be dropped (a notification,
AgentRef). - FRAMES
- What the harness wrote, in a batch of raw lines (
Frames→Items). - INTERRUPT
- Stop the running turn (
Interrupt→Writes). - ITEMS
- From the plugin, without being asked (a notification,
Pushed): what its harness said by a way only the plugin reads (the event stream of a shared server). - LAUNCH
- Stateless: the description of a harness’s launch for an agent (
LaunchParams→Launch). - LET_GO
- The agent was the last in its folder: its shared harness lets that folder go (
AgentRef). - ONE_
SHOT - Stateless: the launch of a one-shot question (
OneShotParams→OneShotLaunch). - ONE_
SHOT_ READ - Stateless: the answer of a one-shot question, from what its process wrote (
OneShotOutput→OneShotText). - OPEN
- A harness shared by agents (
Launch::shared): the agent’s session opened on it (Open→Opened). - PROGRAM_
GONE - Its program ended by itself, for a harness that declares
relaunches(Gone→Items):Item::Relaunchgoes on with another,Item::Stopsays why not, nothing ends the agent’s process. - READ_
USAGE - The quota of the agent’s account, as its process reads it (
AgentRef→Usage). - REMOTE
- Link its session to its vendor’s site, or end the link (
RemoteLink→Asked). - RENAME
- Give its session a title (
SetValue). - SEND_
MESSAGE - A message to send (
SendMessage→Writes). - SET_
EFFORT - SET_
MODE - SET_
MODEL - A change of model, effort or mode (
SetValue→Writes). - STOP_
TASK - Stop one of its background tasks (
Task→Asked). - TASK_
OUTPUT - What a background task wrote so far (
Task→Asked). - TERMINAL_
HOOK - One of its hooks, with what the core answers it (
Hook→HookAnswer). - TERMINAL_
LAUNCH - The launch of the harness’s own interface in a terminal (
TerminalParams→Launch).