AxAgentContextEvent

TypeScript

type AxAgentContextEvent = 
  | {
  actionLogEntryCount: number;
  checkpointActive: boolean;
  effectiveBudgetChars: number;
  fixedPromptChars: number;
  guidanceLogEntryCount: number;
  kind: "budget_check";
  mutablePromptChars: number;
  pressure: AxAgentContextPressure;
  stage: AxAgentContextStage;
  targetPromptChars: number;
  turn: number;
}
  | {
  coveredTurns: number[];
  kind: "checkpoint_created" | "checkpoint_cleared";
  reason: "over_budget" | "under_budget" | "disabled";
  stage: AxAgentContextStage;
  summaryChars?: number;
  turn: number;
}
  | {
  kind: "tombstone_created";
  resolvedByTurn: number;
  source: "deterministic" | "model";
  stage: AxAgentContextStage;
  summaryChars: number;
  turn: number;
}
  | {
  kind: "action_compacted";
  mode: "distill" | "compact";
  originalChars: number;
  reason: "structured_output" | "superseded" | "pressure" | "proactive" | "lean";
  renderedChars: number;
  stage: AxAgentContextStage;
  turn: number;
}
  | {
  domain: "modules" | "skills" | "memories";
  kind: "relevance_ranking";
  shortlist: object[];
  stage: AxAgentContextStage;
  suppressed: boolean;
  taskChars: number;
}
  | {
  fieldName: string;
  kind: "field_auto_promoted";
  originalChars: number;
  promptPreviewChars?: number;
  stage: AxAgentContextStage;
  turn: number;
};

Defined in: https://github.com/ax-llm/ax/blob/e4219a44b7581a55a40d9dd62d2c281062ff078d/src/ax/agent/contextEvents.ts#L5

Type declaration

TypeScript

{
  actionLogEntryCount: number;
  checkpointActive: boolean;
  effectiveBudgetChars: number;
  fixedPromptChars: number;
  guidanceLogEntryCount: number;
  kind: "budget_check";
  mutablePromptChars: number;
  pressure: AxAgentContextPressure;
  stage: AxAgentContextStage;
  targetPromptChars: number;
  turn: number;
}

Name	Type
`actionLogEntryCount`	`number`
`checkpointActive`	`boolean`
`effectiveBudgetChars`	`number`
`fixedPromptChars`	`number`
`guidanceLogEntryCount`	`number`
`kind`	`"budget_check"`
`mutablePromptChars`	`number`
`pressure`	`AxAgentContextPressure`
`stage`	`AxAgentContextStage`
`targetPromptChars`	`number`
`turn`	`number`

TypeScript

{
  coveredTurns: number[];
  kind: "checkpoint_created" | "checkpoint_cleared";
  reason: "over_budget" | "under_budget" | "disabled";
  stage: AxAgentContextStage;
  summaryChars?: number;
  turn: number;
}

Name	Type
`coveredTurns`	`number`[]
`kind`	`"checkpoint_created"` \| `"checkpoint_cleared"`
`reason`	`"over_budget"` \| `"under_budget"` \| `"disabled"`
`stage`	`AxAgentContextStage`
`summaryChars?`	`number`
`turn`	`number`

TypeScript

{
  kind: "tombstone_created";
  resolvedByTurn: number;
  source: "deterministic" | "model";
  stage: AxAgentContextStage;
  summaryChars: number;
  turn: number;
}

Name	Type
`kind`	`"tombstone_created"`
`resolvedByTurn`	`number`
`source`	`"deterministic"` \| `"model"`
`stage`	`AxAgentContextStage`
`summaryChars`	`number`
`turn`	`number`

TypeScript

{
  kind: "action_compacted";
  mode: "distill" | "compact";
  originalChars: number;
  reason: "structured_output" | "superseded" | "pressure" | "proactive" | "lean";
  renderedChars: number;
  stage: AxAgentContextStage;
  turn: number;
}

Name	Type
`kind`	`"action_compacted"`
`mode`	`"distill"` \| `"compact"`
`originalChars`	`number`
`reason`	`"structured_output"` \| `"superseded"` \| `"pressure"` \| `"proactive"` \| `"lean"`
`renderedChars`	`number`
`stage`	`AxAgentContextStage`
`turn`	`number`

TypeScript

{
  domain: "modules" | "skills" | "memories";
  kind: "relevance_ranking";
  shortlist: object[];
  stage: AxAgentContextStage;
  suppressed: boolean;
  taskChars: number;
}

Name	Type	Description
`domain`	`"modules"` \| `"skills"` \| `"memories"`	-
`kind`	`"relevance_ranking"`	-
`shortlist`	`object`[]	Items surfaced to the model, most relevant first ([] if suppressed).
`stage`	`AxAgentContextStage`	-
`suppressed`	`boolean`	True when the low-confidence guard emitted nothing.
`taskChars`	`number`	Length of the ranked task string (not the text — avoids log bloat).

Emitted once per ranked domain per forward when the advisory relevance ranker runs (relevanceRanking plus the domain’s prerequisite: modules need functionDiscovery; skills/memories need their catalogs). Records the shortlist actually surfaced to the model.

To measure whether the hint helps, an observer joins per forward: shortlist.map((s) => s.id) against what the model then loaded — for modules the internal discover calls (onFunctionCall with kind:'internal', name:'discover', args.request) and the module part of external qualifiedNames; for skills onLoadedSkills/used(id); for memories onLoadedMemories/used(id).

TypeScript

{
  fieldName: string;
  kind: "field_auto_promoted";
  originalChars: number;
  promptPreviewChars?: number;
  stage: AxAgentContextStage;
  turn: number;
}

Name	Type	Description
`fieldName`	`string`	-
`kind`	`"field_auto_promoted"`	-
`originalChars`	`number`	-
`promptPreviewChars?`	`number`	Chars kept inline as a preview; undefined => runtime-only.
`stage`	`AxAgentContextStage`	-
`turn`	`number`	-

Emitted once per field per run when autoUpgrade.contextFields keeps an oversized undeclared input value runtime-only. The value stays available in the code runtime as inputs.<fieldName>; the prompt carries a truncated preview (or nothing when promptPreviewChars is undefined) plus a contextMetadata entry.