agentfootprint
Version:
The explainable agent framework — backtrack a wrong answer to the exact context that caused it (evidence, not guesses). Built on footprintjs.
32 lines • 1.35 kB
JavaScript
/**
* toolOutcome — the normalized tool-outcome vocabulary, as a PURE leaf
* (9.34.0).
*
* WHY IT LIVES HERE. A skill-graph route edge can key on what a tool
* DECLARED happened (`onToolStatus`), so this vocabulary is part of the
* graph's own reading of a turn — `InjectionContext.toolResults[].status`
* is typed by it. It used to be declared inside `core/agent/toolEffects.ts`
* (today `semantics/toolEffects.ts`),
* which meant the graph's context type reached into the agent loop's
* tool-effects module for one string union: the loop's vocabulary embedded
* in the graph's contract, pointing the dependency the wrong way.
*
* So the words moved DOWN here — a module with zero imports that both
* sides read. `semantics/toolEffects.ts` re-exports them under the same
* names, so every existing import path is unchanged; that module remains
* the owner of the ENVELOPE grammar (the two effect kinds, the recognizer).
*
* Zone: a zero-import leaf the skill-graph PURE CORE may reach (`PURE_LEAVES`).
* Pinned by `test/injection-engine/skill-graph-fence.test.ts`.
*/
/** The closed set, as data — validators and docs read one list. */
export const TOOL_RESULT_STATUSES = [
'success',
'failure',
'denied',
'invalid',
'partial',
'pending',
'absent',
];
//# sourceMappingURL=toolOutcome.js.map