@tangle-network/agent-runtime 0.94.0 → 0.94.2

package/dist/{types-Cg_teUPO.d.ts → types-Bb-CqxJH.d.ts}

@@ -200,6 +200,9 @@ interface Budget {
 interface Spend {
     iterations: number;
     tokens: LoopTokenUsage;
+    /** Dollar accounting is known unless explicitly false. A false value must not be treated as $0
+     *  when enforcing a dollar-denominated comparison or limit. */
+    usdKnown?: boolean;
     usd: number;
     ms: number;
 }

package/dist/{worktree-fanout-BrB4att3.d.ts → worktree-fanout-DqzIBWMW.d.ts}

@@ -1,10 +1,10 @@
 import { AgentProfile } from '@tangle-network/agent-interface';
 import { AnalystFinding, DefaultVerdict } from '@tangle-network/agent-eval';
-import { d as AgentSpec, e as ExecutorRegistry, B as Budget, A as Agent, f as SpawnJournal, a as ResultBlobStore, g as RootHandle, b as SupervisedResult, N as NodeId, h as Settled, i as Spend, S as Scope, c as Executor } from './types-Cg_teUPO.js';
+import { d as AgentSpec, e as ExecutorRegistry, B as Budget, A as Agent, f as SpawnJournal, a as ResultBlobStore, g as RootHandle, b as SupervisedResult, N as NodeId, h as Settled, i as Spend, S as Scope, c as Executor } from './types-Bb-CqxJH.js';
 import { R as RuntimeHooks, I as Iteration } from './types-DWA64rbJ.js';
 import { BackendType } from '@tangle-network/sandbox';
-import { W as WorktreeHarnessResult, G as GitRunner, a as WorktreeCheckRunner, D as DeliverableSpec } from './completion-gate-BDkEYB_-.js';
-import { L as LocalHarness, r as runLocalHarness } from './local-harness-dcD5WTTr.js';
+import { W as WorktreeHarnessResult, G as GitRunner, a as WorktreeCheckRunner, D as DeliverableSpec } from './completion-gate-Bz7VYs7a.js';
+import { L as LocalHarness, r as runLocalHarness } from './local-harness-C2HDGiOY.js';
 
 /**
  *
@@ -802,12 +802,10 @@ interface CoderCheckConstraints {
  * result onto the `Executor` port (artifact + spend) and owns the teardown point. The §1.5 payload
  * (authored systemPrompt + model) reaches the harness inside the core, not here.
  *
- * Token accounting: a harness CLI does not surface usage, so this executor defaults to
- * `budgetExempt: true` — its spend is NOT metered against the conserved pool and its iterations are
- * EXCLUDED from the equal-k arms by construction (mirrors `cliExecutor`). The exemption is an
- * explicit, documented `budgetExempt` option rather than a buried hardcode: set it `false` ONLY for
- * a harness that genuinely surfaces real token/usd usage to meter into the pool — otherwise the
- * executor would meter a fabricated zero, which the no-silent-zeros rule forbids.
+ * Token accounting: ordinary harness CLI runs remain `budgetExempt`. Reproducible Codex mode
+ * parses the CLI's terminal JSONL usage and is metered by default; an absent usage event fails the
+ * run instead of recording fabricated zero tokens. Codex does not report dollar cost, so that
+ * channel is explicitly marked unknown on the resulting `Spend`.
  *
  * @experimental
  */
@@ -831,6 +829,12 @@ interface WorktreeCliExecutorOptions {
     baseRef?: string;
     /** Wall-clock cap per harness subprocess (ms). Default 5 min (the `runLocalHarness` default). */
     harnessTimeoutMs?: number;
+    /** Run Codex with an ephemeral session, isolated config/instructions, network disabled, and
+     *  JSONL usage capture. Requires `harness: 'codex'`; metered by default. */
+    codexReproducible?: boolean;
+    /** Absolute host paths denied to reproducible Codex (for benchmark answer copies, credentials,
+     *  or other task-specific ambient state). */
+    codexReadDeniedPaths?: ReadonlyArray<string>;
     /**
      * Shell command run in the live worktree to derive the tests-PASS signal (e.g. `pnpm test`).
      * Its exit code becomes `artifact.checks.tests.passed`. Omit to skip (no signal derived).
@@ -850,10 +854,9 @@ interface WorktreeCliExecutorOptions {
      *  outcomes without spawning a real shell. Defaults to a `/bin/sh -c` spawn in the worktree. */
     runCommand?: WorktreeCheckRunner;
     /**
-     * Exclude this leaf's spend from the conserved pool + equal-k arms. Defaults to `true` because a
-     * coding-harness CLI does not surface token usage, so metering it would record a fabricated zero
-     * (the no-silent-zeros rule forbids that). Set `false` ONLY for a harness that surfaces real
-     * token/usd usage worth metering — the executor would then debit the (real) spend it captures.
+     * Exclude this leaf's spend from accounting. Defaults to `true` for ordinary CLI runs and
+     * `false` for `codexReproducible`, which captures real token usage. A metered custom runner must
+     * likewise return `LocalHarnessResult.usage`.
      */
     budgetExempt?: boolean;
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@tangle-network/agent-runtime",
-  "version": "0.94.0",
+  "version": "0.94.2",
   "description": "Shared task-lifecycle skeleton for agents: a recursive loop kernel for chat turns, one-shot tasks, and multi-attempt loops, with trace capture and eval-gated self-improvement. Domain behavior lives in adapters; scoring and ship-gates in @tangle-network/agent-eval.",
   "homepage": "https://github.com/tangle-network/agent-runtime#readme",
   "repository": {