cool-workflow 0.1.94 → 0.1.96

package/dist/workbench.js

@@ -110,23 +110,25 @@ function buildPanels(runner, runId) {
  *  never fabricated. */
 function buildWorkbenchRunView(runner, runId) {
     const id = String(runId || "");
-    let resolved = true;
-    let error;
-    try {
-        runner.loadRun(id);
-    }
-    catch (caught) {
-        resolved = false;
-        error = caught instanceof Error ? caught.message : String(caught);
-    }
-    return {
-        schemaVersion: 1,
-        surface: "workbench",
-        runId: id,
-        resolved,
-        ...(error ? { error } : {}),
-        panels: buildPanels(runner, id)
-    };
+    return runner.loadWithCache((r) => {
+        let resolved = true;
+        let error;
+        try {
+            r.loadRun(id);
+        }
+        catch (caught) {
+            resolved = false;
+            error = caught instanceof Error ? caught.message : String(caught);
+        }
+        return {
+            schemaVersion: 1,
+            surface: "workbench",
+            runId: id,
+            resolved,
+            ...(error ? { error } : {}),
+            panels: buildPanels(r, id)
+        };
+    });
 }
 // ---------------------------------------------------------------------------
 // Cross-run entry (v0.1.28 Run Registry) — composed from already-declared

package/dist/worker-isolation.js

@@ -393,7 +393,31 @@ function recordWorkerRetryAttempt(run, workerId, attempts, reason, options = {})
 function validateWorkerBoundary(run, workerId, options = {}) {
     const scope = requireWorkerScope(run, workerId);
     const rawPath = String(options.path || scope.resultPath);
-    return (0, sandbox_profile_1.validateSandboxWrite)(sandboxPolicyForBoundary(run, scope, options), rawPath, workerId);
+    const policy = sandboxPolicyForBoundary(run, scope, options);
+    const violation = (0, sandbox_profile_1.validateSandboxWrite)(policy, rawPath, workerId);
+    if (!violation) {
+        // Write paths are enforced by CW at this boundary. Command and network limits
+        // are declared in the sandbox policy but enforced by the execution backend
+        // (the host/container runtime). Record the policy split transparently so the
+        // audit trail shows what CW checked vs what was delegated to the host.
+        (0, trust_audit_1.recordTrustAuditEvent)(run, {
+            kind: "worker.sandbox-boundary",
+            decision: "allowed",
+            source: "cw-validated",
+            workerId,
+            taskId: scope.taskId,
+            sandboxProfileId: policy.id,
+            policyRef: `execute=${policy.execute.mode} network=${policy.network.mode} env.inherit=${policy.env.inherit}`,
+            command: policy.execute.mode,
+            networkTarget: policy.network.mode,
+            metadata: {
+                enforced_by_cw: ["write-paths"],
+                delegated_to_host: ["execute", "network", "env"],
+                env_inherit: policy.env.inherit
+            }
+        });
+    }
+    return violation;
 }
 function summarizeWorkers(run) {
     const workers = listWorkerScopes(run);

package/docs/agent-delegation-drive.7.md

@@ -27,6 +27,52 @@ HTTP API. Any API key comes from the agent's *own* inherited env; CW never reads
 or keeps a record of it. Adding a provider SDK to `package.json` would lose the
 neutral-audit moat and is the red line.
 
+## Architecture — the boundary (core ↔ agent backend ↔ wrappers)
+
+The core gives only an INTERFACE: the `agent` execution backend plus a small
+text/process contract. The four vendors (claude / codex / gemini / deepseek)
+live OUTSIDE the core as out-of-process wrapper scripts in `scripts/agents/` —
+pure config, never imported by `src/`, behind the same seam.
+
+```text
+                       user
+                       │  cw -q "…" -codex   (headline shortcut; also -claude/-gemini/-deepseek)
+                       ▼
+┌──────────── CW core  (src/ — zero runtime deps, imports NO model SDK, holds NO key) ───────────┐
+│                                                                                                │
+│  command-surface.ts        agent-config.ts                 execution-backend  ("agent" driver) │
+│  -codex → builtin:codex ─► builtin:<name> ─►               runAgentProcess:                     │
+│                            node <dir>/<name>-agent.js       spawnSync(binary, args, shell:false)│
+│                            {{input}} {{result}}            · inherits env  · captures stdout    │
+│                            (builtin-templates.json: DATA)  · records handle{process}+attestation│
+└───────────────────────────────────────┬───────────────────────────────────────────────────────┘
+     THE SEAM  (text / process contract) │
+       in : argv  {{input}}=worker input.md   {{result}}=worker result.md
+       out: wrapper writes result.md + ONE stdout JSON line {model,usage,result} → parseAgentReport
+                                         │
+   ── red line ──  core never reads a key; each wrapper resolves its OWN key from inherited env
+                                         │
+┌──────── external wrappers  (scripts/agents/*.js — "CONFIG, not a CW runtime dependency") ───────┐
+│   claude-p-agent.js     codex-agent.js      gemini-opencode-agent.js    deepseek-agent.js       │
+│        │                     │                        └──────────┬──────────────┘              │
+│        ▼                     ▼                                   ▼  (3-line shims)              │
+│   claude -p            codex exec                          opencode-agent.js                    │
+│   (Anthropic CLI)      (-c effort, sandbox)               opencode run --model …                │
+│                                                           (deepseek + gemini keys live here)    │
+└─────────────────────────────────────────────────────────────────────────────────────────────────┘
+   Add a vendor = drop a wrapper script + one line in builtin-templates.json — NO core edit.
+```
+
+Each box maps to one source seam: the flag map lives in `command-surface.ts`
+(`-codex` → `builtin:codex`); `agent-config.ts` expands `builtin:<name>` into
+`node <dir>/<name>-agent.js {{input}} {{result}}` by reading
+`builtin-templates.json` (the registry is DATA, not a kernel literal); the
+`agent` driver's `runAgentProcess` does the `spawnSync` and reads back the one
+stdout JSON line via `parseAgentReport`. So the four vendors are **delegated
+agent wrappers, not an event-"hook" system**. The seam is generic: any
+`CW_AGENT_COMMAND="node my-agent.js {{input}} {{result}}"` or a configured HTTP
+endpoint plugs in the same way — the four builtins are just bundled examples.
+
 ## Operator-chosen model is policy; agent-reported model is the attestation
 
 Any model id CW passes **into** the agent invocation (`CW_AGENT_MODEL`
@@ -250,7 +296,22 @@ The built-in templates are:
 claude and codex run their own CLIs; gemini and deepseek route through opencode
 (where their keys live), each proven by a local, deterministic wrapper smoke
 (override the model with CW_GEMINI_MODEL / CW_DEEPSEEK_MODEL). GLM stays an
-external agent command or HTTP endpoint. CW still imports no model SDK.
+external agent command or HTTP endpoint. CW still imports no model SDK. The same
+headline shortcuts pick these builtins on the top-level CLI: `cw -q "..."
+-claude` / `-codex` / `-gemini` / `-deepseek`.
+
+The codex wrapper caps codex's reasoning effort for CW runs so a heavy
+`model_reasoning_effort = "high"` in the user's `~/.codex/config.toml` does not
+make every read/grep turn slow. It passes `codex exec -c
+model_reasoning_effort=<effort>` (default `low`) for THAT run only — the user's
+interactive codex is untouched. Raise it with `CW_CODEX_REASONING_EFFORT`
+(`low` | `medium` | `high`).
+
+When an agent hop fails, CW core keeps only the child's stdout + exit code, so a
+bare `failed (exit 1)` hid the real cause (a relay 5xx, an auth error, a killed
+run). Each wrapper now also drops the failed child's stderr to
+`<run>/workers/<worker>/logs/agent-stderr.log`, so the reason is readable after
+the fact without changing the recorded, byte-stable evidence.
 
 ## Compatibility
 
@@ -344,3 +405,7 @@ The one-command `cw -q` headline now routes the question and defaults the repo t
 0.1.93
 
 0.1.94
+
+0.1.95
+
+0.1.96

package/docs/cli-mcp-parity.7.md

@@ -547,3 +547,7 @@ CLI golden-path fixes: `cw -q "…"` routes the question (was read as an app id
 0.1.93
 
 0.1.94
+
+0.1.95
+
+0.1.96

package/docs/contract-migration-tooling.7.md

@@ -165,3 +165,7 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.1.93
 
 0.1.94
+
+0.1.95
+
+0.1.96

package/docs/control-plane-scheduling.7.md

@@ -149,3 +149,7 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.1.93
 
 0.1.94
+
+0.1.95
+
+0.1.96

package/docs/durable-state-and-locking.7.md

@@ -148,3 +148,7 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.1.93
 
 0.1.94
+
+0.1.95
+
+0.1.96

package/docs/evidence-adoption-reasoning-chain.7.md

@@ -309,3 +309,7 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.1.93
 
 0.1.94
+
+0.1.95
+
+0.1.96

package/docs/execution-backends.7.md

@@ -339,3 +339,7 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.1.93
 
 0.1.94
+
+0.1.95
+
+0.1.96

package/docs/multi-agent-cli-mcp-surface.7.md

@@ -129,6 +129,14 @@ The host surface fails closed when:
 - selection is missing a score or verifier readiness
 - a verifier-gated commit is not ready
 
+## PDCA Blackboard Loop
+
+`pdca-blackboard-loop` is a small workflow app for three-role work over one
+blackboard: planner, builder, and auditor. The host can plan the app, post each
+role result to one board topic through MCP, snapshot after the check step, and
+snapshot again after the next action. Builder, auditor, and next-action tasks
+require grounded evidence, so a verdict with no builder evidence is refused.
+
 ## Smoke Coverage
 
 `test/multi-agent-cli-mcp-surface-smoke.js` covers the full host loop over the
@@ -307,3 +315,7 @@ The host-facing surface tracks the CLI golden-path fixes (`cw -q` routing + repo
 0.1.93
 
 0.1.94
+
+0.1.95
+
+0.1.96

package/docs/multi-agent-eval-replay-harness.7.md

@@ -341,3 +341,7 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.1.93
 
 0.1.94
+
+0.1.95
+
+0.1.96

package/docs/multi-agent-operator-ux.7.md

@@ -353,3 +353,7 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.1.93
 
 0.1.94
+
+0.1.95
+
+0.1.96

package/docs/node-snapshot-diff-replay.7.md

@@ -174,3 +174,7 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.1.93
 
 0.1.94
+
+0.1.95
+
+0.1.96

package/docs/observability-cost-accounting.7.md

@@ -233,3 +233,7 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.1.93
 
 0.1.94
+
+0.1.95
+
+0.1.96

package/docs/project-index.md

@@ -1,15 +1,15 @@
 # Cool Workflow Project Index
 
-Generated from the current repository code on 2026-06-26 by `npm run sync:project-index`.
+Generated from the current repository code on 2026-06-28 by `npm run sync:project-index`.
 
 ## Snapshot
 
 - Package: `cool-workflow`
-- Version: `0.1.94`
+- Version: `0.1.96`
 - Source modules: `68`
-- Workflow apps: `7`
+- Workflow apps: `8`
 - Docs: `53`
-- Smoke tests: `152`
+- Smoke tests: `158`
 - Repository: https://github.com/coo1white/cool-workflow
 
 ## Architecture
@@ -128,6 +128,7 @@ multi-agent host -> topology -> blackboard/coordinator
 | `architecture-review` - Map a repository architecture, assess risks, verify important findings, and synthesize an evidence-backed verdict. | canonical | `repo`, `question`, `invariant`, `focus` | `readonly` | [manifest](../apps/architecture-review/app.json) / [workflow](../apps/architecture-review/workflow.js) |
 | `architecture-review-fast` - Run a shorter architecture review with parallel map and assess phases for faster first results. | canonical | `repo`, `question`, `invariant`, `focus`, `sourceContext`, `sourceContextDigest` | `readonly` | [manifest](../apps/architecture-review-fast/app.json) / [workflow](../apps/architecture-review-fast/workflow.js) |
 | `end-to-end-golden-path` - Deterministic one-worker workflow app for proving the CW integration chain. | userland | `question` | `readonly` | [manifest](../apps/end-to-end-golden-path/app.json) / [workflow](../apps/end-to-end-golden-path/workflow.js) |
+| `pdca-blackboard-loop` - Three agents use one blackboard to plan, build, check, and choose the next step. | example | `goal`, `repo`, `acceptance` | `readonly`, `workspace-write` | [manifest](../apps/pdca-blackboard-loop/app.json) / [workflow](../apps/pdca-blackboard-loop/workflow.js) |
 | `pr-review-fix-ci` - Review a pull request or branch, inspect CI failures, diagnose actionable issues, optionally patch, verify, and summarize with evidence. | canonical | `repo`, `pr`, `branch`, `base`, `ci`, `mode` | `readonly`, `workspace-write` | [manifest](../apps/pr-review-fix-ci/app.json) / [workflow](../apps/pr-review-fix-ci/workflow.js) |
 | `release-cut` - Prepare a release with checklist discipline: version checks, changelog, tests, packaging, release notes, and final verification. | canonical | `repo`, `version`, `previousVersion`, `releaseBranch`, `dryRun` | `readonly`, `workspace-write` | [manifest](../apps/release-cut/app.json) / [workflow](../apps/release-cut/workflow.js) |
 | `research-synthesis` - Split a research question into claims, investigate sources, cross-check evidence, verify claims, and synthesize a concise answer. | canonical | `question`, `source`, `scope`, `freshness` | `readonly`, `locked-down` | [manifest](../apps/research-synthesis/app.json) / [workflow](../apps/research-synthesis/workflow.js) |
@@ -193,6 +194,7 @@ multi-agent host -> topology -> blackboard/coordinator
 
 Smoke tests mirror the public contracts. The high-signal suites are:
 
+- [agent-config-atomic-write-smoke.js](../test/agent-config-atomic-write-smoke.js)
 - [agent-delegation-drive-smoke.js](../test/agent-delegation-drive-smoke.js)
 - [append-run-node-no-realloc-smoke.js](../test/append-run-node-no-realloc-smoke.js)
 - [architecture-review-fast-automation-smoke.js](../test/architecture-review-fast-automation-smoke.js)
@@ -223,6 +225,7 @@ Smoke tests mirror the public contracts. The high-signal suites are:
 - [cli-render-smoke.js](../test/cli-render-smoke.js)
 - [clones-gc-smoke.js](../test/clones-gc-smoke.js)
 - [codex-agent-wrapper-smoke.js](../test/codex-agent-wrapper-smoke.js)
+- [collaboration-ops-unit-smoke.js](../test/collaboration-ops-unit-smoke.js)
 - [concurrency-default-smoke.js](../test/concurrency-default-smoke.js)
 - [concurrent-failure-semantics-smoke.js](../test/concurrent-failure-semantics-smoke.js)
 - [concurrent-workflow-dsl-smoke.js](../test/concurrent-workflow-dsl-smoke.js)
@@ -247,6 +250,7 @@ Smoke tests mirror the public contracts. The high-signal suites are:
 - [execution-backend-agent-smoke.js](../test/execution-backend-agent-smoke.js)
 - [execution-backend-ci-smoke.js](../test/execution-backend-ci-smoke.js)
 - [execution-backends-smoke.js](../test/execution-backends-smoke.js)
+- [feedback-ops-unit-smoke.js](../test/feedback-ops-unit-smoke.js)
 - [freebsd-audit-fixes-smoke.js](../test/freebsd-audit-fixes-smoke.js)
 - [gemini-agent-wrapper-smoke.js](../test/gemini-agent-wrapper-smoke.js)
 - [gemini-opencode-agent-wrapper-smoke.js](../test/gemini-opencode-agent-wrapper-smoke.js)
@@ -256,6 +260,7 @@ Smoke tests mirror the public contracts. The high-signal suites are:
 - [loop-bounded-expansion-smoke.js](../test/loop-bounded-expansion-smoke.js)
 - [mcp-app-surface-smoke.js](../test/mcp-app-surface-smoke.js)
 - [mcp-surface-registry-smoke.js](../test/mcp-surface-registry-smoke.js)
+- [mcp-tool-call-coverage-smoke.js](../test/mcp-tool-call-coverage-smoke.js)
 - [multi-agent-cli-mcp-surface-smoke.js](../test/multi-agent-cli-mcp-surface-smoke.js)
 - [multi-agent-eval-determinism-regression-smoke.js](../test/multi-agent-eval-determinism-regression-smoke.js)
 - [multi-agent-eval-replay-harness-smoke.js](../test/multi-agent-eval-replay-harness-smoke.js)
@@ -277,6 +282,7 @@ Smoke tests mirror the public contracts. The high-signal suites are:
 - [operator-ux-smoke.js](../test/operator-ux-smoke.js)
 - [parallel-onramp-smoke.js](../test/parallel-onramp-smoke.js)
 - [parity-doc-sync-smoke.js](../test/parity-doc-sync-smoke.js)
+- [pdca-blackboard-loop-smoke.js](../test/pdca-blackboard-loop-smoke.js)
 - [pii-redaction-smoke.js](../test/pii-redaction-smoke.js)
 - [pipeline-auto-advance-smoke.js](../test/pipeline-auto-advance-smoke.js)
 - [pipeline-runner-smoke.js](../test/pipeline-runner-smoke.js)
@@ -291,6 +297,7 @@ Smoke tests mirror the public contracts. The high-signal suites are:
 - [readme-trust-claim-smoke.js](../test/readme-trust-claim-smoke.js)
 - [real-execution-backends-smoke.js](../test/real-execution-backends-smoke.js)
 - [registry-corrupt-fail-closed-smoke.js](../test/registry-corrupt-fail-closed-smoke.js)
+- [release-check-skip-smoke.js](../test/release-check-skip-smoke.js)
 - [release-flow-smoke.js](../test/release-flow-smoke.js)
 - [release-gate-smoke.js](../test/release-gate-smoke.js)
 - [release-tooling-smoke.js](../test/release-tooling-smoke.js)

package/docs/real-execution-backends.7.md

@@ -181,3 +181,7 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.1.93
 
 0.1.94
+
+0.1.95
+
+0.1.96

package/docs/release-and-migration.7.md

@@ -321,3 +321,7 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.1.93
 
 0.1.94
+
+0.1.95
+
+0.1.96

package/docs/release-tooling.7.md

@@ -85,6 +85,22 @@ The dogfood release smoke and the architecture-review dogfood smoke are separate
 test files. The split keeps the same release and agent-drive proof, but lets
 `test:ci` schedule the two long checks in parallel.
 
+## PR CI Merge Notes
+
+When a PR is ready, list open PRs by creation time and merge the oldest ready
+one first. After each merge, check the next PR again; a new main commit may make
+it need a rebase.
+
+If main moved after the PR branch was made, replay the PR commits on top of the
+new main in a clean worktree. Keep local unrelated changes out of the rebase and
+out of the PR.
+
+Treat CI as the source of truth for what blocks the merge. Read the failed step
+first, fix only that drift, then push again. A common drift is the npm README:
+when `readme-sync-smoke.js` says the package README is stale, run
+`npm run sync:readme`, add only `plugins/cool-workflow/README.md`, and let CI run
+again.
+
 ## Boundary
 
 Release Tooling touches only the build/release surfaces. It adds no runtime
@@ -268,3 +284,7 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.1.93
 
 0.1.94
+
+0.1.95
+
+0.1.96

package/docs/routines.md

@@ -64,6 +64,36 @@ report path and digest.
 The `--metrics` flag is not required and it gives back foreground time used plus
 agent-spawn and result-cache-hit counts for the fast run.
 
+## PDCA Blackboard Development Lessons
+
+When a task asks for agents to work together, first try the parts CW already
+has:
+
+- workflow apps give the work shape
+- worker output gives checked facts
+- the blackboard gives shared state
+- MCP gives tool access to the same state
+- smoke tests prove the loop
+
+Do not make a new MCP server when the existing server can show the same run
+state. Add a workflow app first, then prove the app with one smoke that uses
+both CLI and MCP.
+
+For a three-agent loop, keep the order plain:
+
+```text
+plan -> build -> audit -> next action
+```
+
+Each agent should write one blackboard message and, when there is a result
+file, one artifact ref. Take a snapshot after the audit and after the next
+action. If audit evidence is missing, let the worker evidence gate refuse the
+result instead of adding a new policy layer.
+
+Before a PR, base the branch on `origin/main`, run the generated-doc checks,
+and sync generated docs when the gate says they are stale. This keeps unrelated
+local commits and generated README/index drift out of the work.
+
 ## Boundary
 
 CW v0.1.1 does not give managed cloud infrastructure. It gives a local

package/docs/run-registry-control-plane.7.md

@@ -464,3 +464,7 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.1.93
 
 0.1.94
+
+0.1.95
+
+0.1.96

package/docs/run-retention-reclamation.7.md

@@ -232,3 +232,7 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.1.93
 
 0.1.94
+
+0.1.95
+
+0.1.96

package/docs/sandbox-profiles.7.md

@@ -26,6 +26,21 @@ normalization, worker result acceptance, and durable feedback for denied worker
 output. The agent host has to do OS-level file access, process execution,
 network access, and environment filtering.
 
+**IMPORTANT**: Under the default `node` backend, a sandbox profile's
+`execute`, `network`, and `env` policy is **attested, not enforced**. CW
+validates the policy, records it in the worker manifest, and attests that
+these limits were declared — but the actual enforcement of command execution
+restrictions, network isolation, and environment variable filtering is
+DELEGATED to the host runtime. For full enforcement, use the `container`
+backend (`--backend container`) with Docker/Podman, or apply OS-level
+sandboxing to the agent process. Without OS enforcement, a worker under a
+`locked-down` profile can still run arbitrary commands and access the network.
+
+CW also now (v0.1.95) applies `buildChildEnv(policy)` as a baseline for agent
+spawns — only `PATH`, `HOME`, explicit `expose` entries, and well-known
+`CW_*` + LLM provider API key environment variables pass through. The
+operator's other process environment is not inherited by default.
+
 The design goal is simple:
 
 ```text

package/docs/state-explosion-management.7.md

@@ -310,3 +310,7 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.1.93
 
 0.1.94
+
+0.1.95
+
+0.1.96

package/docs/team-collaboration.7.md

@@ -246,3 +246,7 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.1.93
 
 0.1.94
+
+0.1.95
+
+0.1.96

package/docs/web-desktop-workbench.7.md

@@ -254,3 +254,7 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.1.93
 
 0.1.94
+
+0.1.95
+
+0.1.96

package/manifest/plugin.manifest.json

@@ -2,7 +2,7 @@
   "_comment": "SINGLE SOURCE OF TRUTH for every vendor manifest. Edit THIS file, then run `npm run gen:manifests`. Do NOT hand-edit the generated vendor manifests (.claude-plugin/, .codex-plugin/, .agents/, .mcp.json) — `npm run gen:manifests -- --check` (run by release:check) will fail if they drift from this source.",
   "identity": {
     "name": "cool-workflow",
-    "version": "0.1.94",
+    "version": "0.1.96",
     "license": "BSD-2-Clause",
     "homepage": "https://github.com/coo1white/cool-workflow",
     "author": {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "cool-workflow",
-  "version": "0.1.94",
+  "version": "0.1.96",
   "bin": {
     "cool-workflow": "scripts/cw.js",
     "cw": "scripts/cw.js"
@@ -14,7 +14,7 @@
   "bugs": {
     "url": "https://github.com/coo1white/cool-workflow/issues"
   },
-  "description": "A workflow control plane and run-time you are able to check: it sends out jobs in TypeScript, makes certain of work against facts before it goes through, puts state into fixed records, orders jobs by time, runs jobs again and again, gets a group of agents to do their parts together, and talks MCP. It gives the doing of the work to outside agents — it never runs the models itself.",
+  "description": "A workflow control plane and run-time you are able to check: it sends out jobs in TypeScript, makes certain of work against facts before it goes through, puts state into fixed records, orders jobs by time, runs jobs again and again, gets a group of agents to do their parts together, and talks MCP. It gives the doing of the work to outside agents \u2014 it never runs the models itself.",
   "type": "commonjs",
   "engines": {
     "node": ">=18"
@@ -62,9 +62,11 @@
     "onramp:check": "node scripts/onramp-check.js --check",
     "version:sync": "node scripts/version-sync-check.js",
     "release:check": "node scripts/release-check.js",
-    "test": "node dist/cli.js version > /dev/null && node test/run-all.js",
+    "test": "node dist/cli.js version > /dev/null && node test/run-all.js --fast --sample 35",
+    "test:full": "node dist/cli.js version > /dev/null && node test/run-all.js --sample 55",
+    "test:gate": "node dist/cli.js version > /dev/null && node test/run-all.js",
     "test:fast": "npm run build --if-present && node dist/cli.js version > /dev/null && node test/run-all.js --concurrency auto",
-    "test:ci": "node dist/cli.js version > /dev/null && node test/run-all.js --concurrency auto",
+    "test:ci": "npm run build && node dist/cli.js version > /dev/null && node test/run-all.js --sample 55",
     "test:coverage": "node dist/cli.js version > /dev/null && node scripts/coverage-gate.js --concurrency auto",
     "eval:replay": "tsc -p tsconfig.json && node test/multi-agent-eval-replay-harness-smoke.js",
     "ci": "npm run build && npm run check && npm run test && npm run release:check",
@@ -85,5 +87,6 @@
     "llm-pipeline",
     "claude",
     "delegation"
-  ]
+  ],
+  "mcpName": "io.github.coo1white/cool-workflow"
 }

package/scripts/agents/agent-adapter-core.js

@@ -2,6 +2,7 @@
 "use strict";
 
 const fs = require("node:fs");
+const path = require("node:path");
 
 const RESULT_CONTRACT = `
 === HOW TO RETURN YOUR ANSWER (overrides any 'write to result.md' instruction above) ===
@@ -448,6 +449,25 @@ function emitReport(model, usage, resultText) {
   process.stdout.write(JSON.stringify({ model, usage, result: resultText }));
 }
 
+// Drop the failed agent's stderr beside the worker's result.md so a `failed (exit
+// 1)` is readable AFTER the fact. CW core keeps only the child's stdout + exit
+// code (byte-stable evidence), so without this the real reason — a relay 5xx, an
+// auth error, a killed run — is lost. The worker dir already has a `logs/` folder
+// (src/worker-isolation.ts); resultPath is `<workerDir>/result.md`, so its
+// `logs/agent-stderr.log` sibling is the natural home. Advisory only: never throws,
+// never changes the exit code or the recorded evidence.
+function persistStderr(resultPath, text) {
+  const t = String(text || "").trim();
+  if (!t || !resultPath) return;
+  try {
+    const dir = path.join(path.dirname(resultPath), "logs");
+    fs.mkdirSync(dir, { recursive: true });
+    fs.writeFileSync(path.join(dir, "agent-stderr.log"), `${t}\n`, "utf8");
+  } catch {
+    /* advisory only — diagnostics must never break the run */
+  }
+}
+
 module.exports = {
   RESULT_CONTRACT,
   buildPrompt,
@@ -461,5 +481,6 @@ module.exports = {
   parseJsonLines,
   flushJsonLines,
   writeResult,
-  emitReport
+  emitReport,
+  persistStderr // save a failed agent's stderr to <workerDir>/logs/agent-stderr.log (shared by all wrappers)
 };