/**
 * Default per-field character cap when serializing content text. Bounds
 * the contribution of any single content field so a giant tool_call args
 * payload cannot dominate the BM25 corpus (or blow up an embedding pipeline
 * by tokenizing megabytes of code). 2048 chars is empirically chosen to
 * preserve most natural-language content while truncating large blobs.
 */
export declare const DEFAULT_FIELD_CAP = 2048;
export interface ExtractIndexableTextOptions {
    /** Maximum characters drawn from any single content field (default 2048). */
    fieldCap?: number;
}
/**
 * Observation content per the EmitInput Zod schema in services/atrib-emit.
 * Indexable fields: `what`, `why_noted`, `intent`, `rationale`, `topics`.
 * `informed_by` is a list of sha256 record_hash refs (not human-readable
 * text) and is omitted from the indexable surface.
 */
export interface ObservationContent {
    what?: unknown;
    why_noted?: unknown;
    intent?: unknown;
    rationale?: unknown;
    topics?: unknown;
    informed_by?: unknown;
    source?: unknown;
    span_kind?: unknown;
    span_name?: unknown;
    trace_id?: unknown;
    span_id?: unknown;
    tool_name?: unknown;
    input?: unknown;
    output?: unknown;
    agent_name?: unknown;
    model_name?: unknown;
    prompt?: unknown;
    prompt_messages?: unknown;
    prompt_tools?: unknown;
    prompt_tool_choice?: unknown;
    prompt_template?: unknown;
    prompt_template_variables?: unknown;
    prompt_version?: unknown;
    prompt_id?: unknown;
    prompt_url?: unknown;
    invocation_parameters?: unknown;
    usage_details?: unknown;
    cost_details?: unknown;
    score_details?: unknown;
    metadata?: unknown;
}
/**
 * Annotation content per services/atrib-annotate. The agent (or a separate
 * curator) tags a past record with `importance` + `summary` + `topics`.
 * `annotates` is a sha256 ref to the target record and is omitted from the
 * indexable surface.
 */
export interface AnnotationContent {
    annotates?: unknown;
    importance?: unknown;
    summary?: unknown;
    topics?: unknown;
}
/**
 * Revision content per services/atrib-revise. The agent supersedes a prior
 * position with a stated reason; indexable fields are `prior_position`,
 * `new_position`, `reason`, `topics`. `revises` is a sha256 ref and is
 * omitted from the indexable surface.
 */
export interface RevisionContent {
    revises?: unknown;
    prior_position?: unknown;
    new_position?: unknown;
    reason?: unknown;
    topics?: unknown;
}
/**
 * Tool-call content per @atrib/mcp wrapper. The wrapper signs each MCP
 * tool invocation; sidecar typically carries tool_name + args + result.
 * Args and result are accepted under several legacy field names because
 * different host frameworks (Claude Agent SDK, Vercel AI SDK, LangChain)
 * historically wrote different keys. JSON-stringified args/result are
 * truncated to the field cap per `ExtractIndexableTextOptions.fieldCap`.
 */
export interface ToolCallContent {
    tool_name?: unknown;
    args?: unknown;
    input?: unknown;
    arguments?: unknown;
    result?: unknown;
    output?: unknown;
    response?: unknown;
}
/**
 * Transaction content per payment-protocol adapters (x402, ACP, UCP, etc.).
 * Counterparty + memo are typically the most indexable fields; amounts are
 * numeric and don't tokenize usefully so they're omitted.
 */
export interface TransactionContent {
    counterparty?: unknown;
    recipient?: unknown;
    merchant?: unknown;
    to?: unknown;
    memo?: unknown;
    description?: unknown;
    protocol?: unknown;
    via?: unknown;
}
/**
 * Directory-anchor content per directory operators (spec §6.2.4). Mostly
 * structural (tree_root, epoch_id); indexable text is minimal but preserved
 * for completeness so anchors are at least surface-able by tree_root.
 */
export interface DirectoryAnchorContent {
    tree_root?: unknown;
    root?: unknown;
    epoch_id?: unknown;
}
/**
 * Extract indexable text from a sidecar content payload.
 *
 * Dispatches on the event_type URI to the per-event_type extractor for
 * normative event types; falls back to a recursive string-walk for
 * extension URIs. Returns `""` when content is undefined / not an object /
 * structurally malformed (silent failure per §5.8 degradation).
 *
 * Suitable for tokenization (BM25, embeddings) or for display synthesis.
 * Per-field length caps prevent giant tool_call payloads from dominating
 * the corpus.
 *
 * @param eventTypeUri - the record's `event_type` field (absolute URI per §1.2.4)
 * @param content      - the D062 sidecar `_local.content` payload (typed
 *                       `unknown` because shape varies per event_type;
 *                       runtime shape-checking handles malformed input)
 * @param opts.fieldCap - max characters per single content field (default 2048)
 */
export declare function extractIndexableText(eventTypeUri: string, content: unknown, opts?: ExtractIndexableTextOptions): string;
export declare function extractObservationText(c: ObservationContent, cap: number): string;
export declare function extractAnnotationText(c: AnnotationContent, cap: number): string;
export declare function extractRevisionText(c: RevisionContent, cap: number): string;
export declare function extractToolCallText(c: ToolCallContent, cap: number): string;
export declare function extractTransactionText(c: TransactionContent, cap: number): string;
export declare function extractDirectoryAnchorText(c: DirectoryAnchorContent, cap: number): string;
//# sourceMappingURL=content-shapes.d.ts.map