Garry's Opinionated OpenClaw/Hermes Agent Brain
TypeScript
/**
* BrainBench OpenClaw adapter — seam: 'production'.
*
* Drives the exact pipeline the shipped OpenClaw context engine runs per turn
* (src/core/context-engine.ts → buildReflexAddition), with production
* defaults: 3-pointer budget, prior-context suppression, markdown pointer
* block, PLUS the volunteer arm (2026-08 fix wave — Arm 2): a 4-turn window
* (DEFAULT_WINDOW_TURNS parity) feeds the SAME volunteerStage primitive
* production runs (via volunteerContext, the engine-bound wrapper), and the
* wire text composes through the SAME renderReflexAddition. What this row
* scores is what an OpenClaw user's reflex actually does — the volunteer
* logic cannot drift between bench and production because both consume one
* primitive (eng-review E1; decision 13's orchestration differences — config
* gate, heartbeat, timeout — stay deliberately ungraded, see BRAINBENCH.md).
*/
import type { BrainEngine } from '../../../core/engine.ts';
import { DEFAULT_MAX_POINTERS } from '../../../core/context/retrieval-reflex.ts';
import { volunteerContext, VOLUNTEER_DEFAULT_MAX_PAGES } from '../../../core/context/volunteer.ts';
import { DEFAULT_WINDOW_TURNS, renderReflexAddition } from '../../../core/context/reflex.ts';
import type { WindowTurn } from '../../../core/context/entity-salience.ts';
import type {
AdapterFixtureView,
HarnessAdapter,
HarnessTurnResult,
PublicTurn,
} from '../types.ts';
import { runReflexPipeline, estimateTokens } from './shared.ts';
export class OpenClawAdapter implements HarnessAdapter {
readonly name = 'openclaw' as const;
readonly seam = 'production' as const;
private engine: BrainEngine | null = null;
private sourceId = 'default';
private window: WindowTurn[] = [];
async beginConversation(engine: BrainEngine, fixture: AdapterFixtureView): Promise<void> {
this.engine = engine;
this.sourceId = fixture.active_source;
this.window = [];
}
async replayTurn(turn: PublicTurn, priorContextText: string): Promise<HarnessTurnResult> {
if (!this.engine) throw new Error('openclaw adapter: beginConversation not called');
const started = performance.now();
this.window.push({ role: turn.role, text: turn.text });
if (this.window.length > DEFAULT_WINDOW_TURNS) {
this.window = this.window.slice(-DEFAULT_WINDOW_TURNS);
}
// Arm 1: pointer budget — unchanged row semantics (E3 pins codex and
// claude-code rows byte-identical; this arm keeps openclaw's pointer
// behavior identical too, the volunteer arm is strictly additive).
const block = await runReflexPipeline(this.engine, this.sourceId, turn, priorContextText, {
maxPointers: DEFAULT_MAX_POINTERS,
suppression: 'prior-context',
});
const pointers = block?.pointers ?? [];
// Arm 2: the production volunteer primitive, same gate constants.
const volunteered = await volunteerContext(this.engine, this.window, {
sourceIds: [this.sourceId],
priorContext: priorContextText,
excludeSlugs: new Set(pointers.map((p) => p.slug)),
maxPages: VOLUNTEER_DEFAULT_MAX_PAGES,
});
const wireText = renderReflexAddition(block?.text ?? null, volunteered);
const latencyMs = performance.now() - started;
return {
injectedText: wireText,
injectedSlugs: [...pointers.map((p) => p.slug), ...volunteered.map((v) => v.slug)],
pointers,
injectedTokens: estimateTokens(wireText),
latencyMs,
};
}
async endConversation(): Promise<void> {
this.engine = null;
this.window = [];
}
}