refactor(bench): use runtime candidate workspaces

bench/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@tangle-network/agent-bench",
-  "version": "0.2.2",
+  "version": "0.3.0",
   "type": "module",
   "description": "The unified benchmark suite for agent-runtime agents: 31 adapters (commit0, enterpriseops-gym, ragbench, crag, nomiracl, open-rag-bench, t2-ragbench, tau3-banking, bfcl, finresearchbench, …) behind one resolveAdapter registry, each with a real judge or fail-loud unsupported scorer. Score any profile/skill/prompt change against them. Map: bench/HARNESS.md.",
   "repository": {
@@ -27,9 +27,9 @@
     "verify:pier": "tsx scripts/verify-pier-pair.mts"
   },
   "dependencies": {
-    "@tangle-network/agent-eval": "^0.114.0",
+    "@tangle-network/agent-eval": "^0.115.2",
     "@tangle-network/agent-interface": "^0.25.0",
-    "@tangle-network/agent-runtime": "^0.92.0",
+    "@tangle-network/agent-runtime": "0.94.6",
     "@tangle-network/sandbox": "^0.10.3"
   },
   "devDependencies": {
@@ -39,7 +39,7 @@
   },
   "pnpm": {
     "overrides": {
-      "@tangle-network/agent-eval": "0.114.0"
+      "@tangle-network/agent-eval": "0.115.2"
     }
   },
   "files": [

bench/scripts/verify-pier-agent.mts

@@ -23,6 +23,7 @@ import type {
   Sha256Digest,
 } from '@tangle-network/agent-interface'
 import {
+  createAgentCandidateWorkspacePort,
   disposePreparedAgentCandidateExecution,
   FileAgentCandidateExecutionClaimStore,
   prepareAgentCandidateExecution,
@@ -36,7 +37,6 @@ import {
 import { executePreparedPierCandidate } from '../src/pier-agent'
 import { createPierResultGrader } from '../src/pier-result-grader'
 import { FilePierCandidateTrialController } from '../src/pier-trial-controller'
-import { materializePierWorkspaceArchive } from '../src/pier-workspace-archive'
 
 const pinnedPierCommit = 'e69a20e4e0ac073ec71fde0274bab3d9f40bac87'
 const pinnedPierVersion = '0.3.0'
@@ -441,6 +441,7 @@ ${proofArm === 'success' ? "(task / 'src/status.txt').write_text('ready\\nowner=
       maxCostUsd: 0,
     },
   }
+  const workspaces = createAgentCandidateWorkspacePort()
   const ports: AgentCandidateExecutionPorts = {
     artifacts: {
       read: async () => {
@@ -449,9 +450,14 @@ ${proofArm === 'success' ? "(task / 'src/status.txt').write_text('ready\\nowner=
     },
     repositories: { resolve: async () => taskRoot },
     workspaces: {
-      materialize: async ({ snapshot, archive, destination }) => {
+      materialize: async ({ role, snapshot, archive, destination }) => {
         if (destination === taskRoot || destination === candidateRoot) return
-        await materializePierWorkspaceArchive({ archive, expected: snapshot.material, destination })
+        await workspaces.materialize({
+          role,
+          snapshot,
+          archive,
+          destination,
+        })
       },
     },
     containers: { resolve: async () => container },

bench/src/index.ts

@@ -73,10 +73,3 @@ export {
   type PierDockerConnection,
 } from './pier-trial-controller'
 export { createPierResultGrader } from './pier-result-grader'
-export {
-  createPierWorkspaceArchive,
-  materializePierWorkspaceArchive,
-  type PierWorkspaceArchiveFile,
-  type PierWorkspaceArchiveV1,
-  type PierWorkspaceFile,
-} from './pier-workspace-archive'

bench/src/pier-agent.ts

@@ -592,6 +592,10 @@ export async function executePreparedPierCandidate(
         baseCommit: repository.baseCommit,
         baseTree: repository.baseTree,
         patch: result.taskPatch,
+        artifactPersistence: {
+          executionId: request.executionId,
+          outputArtifacts: options.outputArtifacts,
+        },
       })
       return { stopped: true, taskOutcome }
     },