cowork-harness 0.4.0 → 0.4.1

package/CHANGELOG.md

@@ -6,6 +6,18 @@ All notable changes to this project are documented here. The format is based on
 
 ## [Unreleased]
 
+## [0.4.1] — 2026-06-18
+
+### Fixed
+
+- **Agent-binary newest-staged fallback now applies on the real runtime paths** (container / hostloop, and
+  thus `skill` / `run` / `chat`), not just `sync`/tests. `resolveAgentBinary` had two private duplicates
+  (`container.ts`, `hostloop.ts`) **without** the 0.4.0 fallback, so a host with a newer staged
+  `claude-code-vm/<ver>` than the baseline expects still hard-failed with "Staged agent binary not found".
+  The duplicates were consolidated into the single exported resolver; a host that has staged a newer build
+  now falls back to it (with a warning) instead of failing. A structural test + CI guard prevent the
+  resolver from being re-duplicated.
+
 ## [0.4.0] — 2026-06-18
 
 The parsing/validation hardening + safety release: a current-tree code-review sweep plus fidelity and

package/dist/runtime/container.js

@@ -1,9 +1,7 @@
 import { spawn } from "node:child_process";
-import { existsSync } from "node:fs";
 import { resolve, join } from "node:path";
-import { homedir } from "node:os";
 import { DEFAULT_MAX_THINKING_TOKENS } from "../types.js";
-import { resolveMounts } from "../baseline.js";
+import { resolveMounts, resolveAgentBinary } from "../baseline.js";
 import { agentArgs, spawnEnv, dockerRunArgv, resolveMaxThinkingTokens } from "./argv.js";
 import { runtimeAuthEnv } from "./host-env.js";
 import { stageWorkspace } from "./stage.js";
@@ -60,18 +58,3 @@ export function spawnContainer(scenario, baseline, plan, outDir, sessionId, opts
     });
     return spawn(runner, dockerArgs, { stdio: ["pipe", "pipe", "pipe"] });
 }
-/** Resolve the staged Linux agent binary on the host (override: COWORK_AGENT_BINARY). */
-function resolveAgentBinary(baseline) {
-    const override = process.env.COWORK_AGENT_BINARY;
-    if (override) {
-        if (!existsSync(override))
-            throw new Error(`COWORK_AGENT_BINARY not found: ${override}`);
-        return resolve(override);
-    }
-    const staged = (baseline.agentBinary?.stagedPath ?? "").replace(/^~(?=$|\/)/, homedir());
-    if (!staged || !existsSync(staged)) {
-        throw new Error(`Staged agent binary not found at "${staged}". It is extracted from your Claude Desktop install ` +
-            `(claude-code-vm/<ver>/claude). Open Cowork once to stage it, or set COWORK_AGENT_BINARY to its path.`);
-    }
-    return resolve(staged);
-}

package/dist/runtime/hostloop.js

@@ -1,11 +1,10 @@
 import { warn } from "../io.js";
 import { spawn } from "node:child_process";
-import { existsSync, readFileSync } from "node:fs";
+import { readFileSync } from "node:fs";
 import { resolve, join } from "node:path";
-import { homedir } from "node:os";
 import { fileURLToPath } from "node:url";
 import { DEFAULT_MAX_THINKING_TOKENS } from "../types.js";
-import { resolveMounts } from "../baseline.js";
+import { resolveMounts, resolveAgentBinary } from "../baseline.js";
 import { makeWorkspaceHandler } from "../hostloop/workspace-handler.js";
 import { agentArgs, spawnEnv, dockerRunArgv, resolveMaxThinkingTokens } from "./argv.js";
 import { runtimeAuthEnv } from "./host-env.js";
@@ -116,16 +115,3 @@ function hostLoopShellSection(vmMnt, appVersion) {
         .join(vmMnt)
         .trim();
 }
-function resolveAgentBinary(baseline) {
-    const override = process.env.COWORK_AGENT_BINARY;
-    if (override) {
-        if (!existsSync(override))
-            throw new Error(`COWORK_AGENT_BINARY not found: ${override}`);
-        return resolve(override);
-    }
-    const staged = (baseline.agentBinary?.stagedPath ?? "").replace(/^~(?=$|\/)/, homedir());
-    if (!staged || !existsSync(staged)) {
-        throw new Error(`Staged agent binary not found at "${staged}". Open Cowork once to stage it, or set COWORK_AGENT_BINARY.`);
-    }
-    return resolve(staged);
-}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "cowork-harness",
-  "version": "0.4.0",
+  "version": "0.4.1",
   "description": "Scriptable, CI-friendly harness for Claude Cowork's runtime contract for testing skills across scenarios — same agent, mounts, egress allowlist, permission protocol, and sandbox limitations.",
   "license": "MIT",
   "type": "module",