cool-workflow 0.2.5 → 0.2.6

package/dist/wiring/capability-table/parity.js

@@ -413,20 +413,25 @@ function lintRegistry() {
 }
 /**
  * Compare the declared registry against the ACTUAL surfaces and report every
- * fail-closed gap. `mcpTools` is the live `tools/list` result; `cliTokens` is the
- * set of `case "<token>"` strings parsed from the CLI source.
+ * fail-closed gap. `mcpTools` is the live `tools/list` result. `cliTokens` is
+ * optional test input for an independent token source; the table-driven live
+ * gate leaves it out and probes dispatcher reachability instead.
  */
 function buildParityReport(input) {
     const declaredTools = new Set((0, registry_core_1.declaredMcpTools)());
     const actualTools = new Set(input.mcpTools);
     const declaredTokens = new Set(declaredCliTokens());
-    const actualTokens = new Set(input.cliTokens);
+    const actualTokens = new Set(input.cliTokens || []);
     const declaredHelpTokens = new Set(declaredCliHelpTokens());
     const actualHelpTokens = new Set(input.helpTokens || []);
     const missingMcpTools = [...declaredTools].filter((tool) => !actualTools.has(tool)).sort();
     const undeclaredMcpTools = [...actualTools].filter((tool) => !declaredTools.has(tool)).sort();
-    const missingCliTokens = [...declaredTokens].filter((token) => !actualTokens.has(token)).sort();
-    const undeclaredCliTokens = [...actualTokens].filter((token) => !declaredTokens.has(token)).sort();
+    const missingCliTokens = input.cliTokens
+        ? [...declaredTokens].filter((token) => !actualTokens.has(token)).sort()
+        : [];
+    const undeclaredCliTokens = input.cliTokens
+        ? [...actualTokens].filter((token) => !declaredTokens.has(token)).sort()
+        : [];
     const helpMissingCliTokens = input.helpTokens
         ? [...declaredHelpTokens].filter((token) => !actualHelpTokens.has(token)).sort()
         : [];

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

@@ -438,3 +438,5 @@ The one-command `cw -q` headline now routes the question and defaults the repo t
 0.2.4
 
 0.2.5
+
+0.2.6

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

@@ -97,6 +97,20 @@ and any other request falls back to the newest supported entry (see
 `SUPPORTED_PROTOCOL_VERSIONS` in `src/mcp/server.ts`). With today's
 one-entry list the reply bytes are unchanged.
 
+## MCP Tool Authority
+
+`CW_MCP_ENABLED_TOOLS` and `CW_MCP_DISABLED_TOOLS` are optional comma lists of
+exact MCP tool names. Space at either end of a name is taken out, and a second
+name has no effect. The enabled list is an allowlist. The disabled list is then
+applied, so deny has the last word. With both values unset, the server keeps its
+full `tools/list` output and present tool access.
+
+An empty name or a name not in the live registry stops server start with a
+stderr diagnostic and a non-zero exit. Under a good policy, `tools/list` keeps
+the registry order and shows only allowed tools. A call to a filtered tool gives
+the normal MCP `isError: true` result and does not start its handler. This is a
+server control, not an OS sandbox or an approval system.
+
 ## The Parity Matrix
 
 The matrix below is made from the live registry — one row per capability,
@@ -463,11 +477,11 @@ The shared arg helpers (`required`, `optionalArg`, `wantsJson`) live in
 `src/core/util/cli-args.ts`, a pure module used by the CLI and MCP sides of
 every wiring slice; they were moved out of `cli/io.ts` because the purity
 gate's layer rule says `wiring/` may not take imports from `cli/`. The parity
-scanner works the same way: `cliDispatchTokens()` gets the live token list
-straight from the registry's CLI capabilities (`cli.caseTokens` when a row
-names more than one first token, else `cli.path`), then makes sure each
-declared path resolves through the live dispatcher — there is no grep over
-built dispatch source.
+gate does not use a token list made from the registry as proof of that same
+registry. It runs real `cw help`, checks every declared CLI path through the
+dispatcher lookup used by `dispatch(args)`, reads MCP `tools/list` from a live
+server, and runs CLI-process to MCP payload probes. There is no grep over built
+dispatch source.
 
 `test/cli-mcp-parity-smoke.js` proves the contract from end to end. It checks
 registry ⇄ CLI ⇄ MCP coverage (every declared capability is found on its declared
@@ -668,3 +682,5 @@ CLI golden-path fixes: `cw -q "…"` routes the question (was read as an app id
 0.2.4
 
 0.2.5
+
+0.2.6

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

@@ -185,3 +185,5 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.2.4
 
 0.2.5
+
+0.2.6

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

@@ -203,3 +203,5 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.2.4
 
 0.2.5
+
+0.2.6

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

@@ -188,3 +188,5 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.2.4
 
 0.2.5
+
+0.2.6

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

@@ -329,3 +329,5 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.2.4
 
 0.2.5
+
+0.2.6

package/docs/execution-backends.7.md

@@ -359,3 +359,5 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.2.4
 
 0.2.5
+
+0.2.6

package/docs/mcp-app-surface.7.md

@@ -21,6 +21,27 @@ The bridge keeps to CW's base-system rules:
   protocol errors to the model still lets it read the message and try
   again
 
+## Control and Tool Processes
+
+The MCP parent process reads JSON-RPC, writes JSON-RPC, and answers `ping`.
+One child process runs `tools/call` work in order. A file lock wait or a long
+agent call in that child does not stop the parent from answering `ping`.
+
+The child sends back the same JSON text that the old in-process tool call gave
+back. Tool names, input names, output shapes, run files, and CLI parity stay
+the same.
+
+If the child stops during a tool call, CW gives an `isError: true` result for
+that call. CW does not run the call again, because it may have written part of
+a run. Read the saved run state before you try the call again. The next tool
+call starts a new child process.
+
+When stdin ends, CW keeps the ordered tool queue to its end and then stops the
+child. This lets a client send a last batch and close stdin without losing its
+replies. On `SIGINT` or `SIGTERM`, CW stops the child at once and then ends by
+the same signal. A tool waiting on a file lock cannot wake later and write
+state after its parent has ended.
+
 ## App Run Flow
 
 Use `cw_app_list`, `cw_app_show`, and `cw_app_validate` to look at app

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

@@ -335,3 +335,5 @@ The host-facing surface tracks the CLI golden-path fixes (`cw -q` routing + repo
 0.2.4
 
 0.2.5
+
+0.2.6

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

@@ -361,3 +361,5 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.2.4
 
 0.2.5
+
+0.2.6

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

@@ -373,3 +373,5 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.2.4
 
 0.2.5
+
+0.2.6

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

@@ -194,3 +194,5 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.2.4
 
 0.2.5
+
+0.2.6

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

@@ -254,3 +254,5 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.2.4
 
 0.2.5
+
+0.2.6

package/docs/project-index.md

@@ -1,15 +1,15 @@
 # Cool Workflow Project Index
 
-Generated from the current repository code on 2026-07-13 by `npm run sync:project-index`.
+Generated from the current repository code on 2026-07-14 by `npm run sync:project-index`.
 
 ## Snapshot
 
 - Package: `cool-workflow`
-- Version: `0.2.5`
-- Source modules: `151`
+- Version: `0.2.6`
+- Source modules: `153`
 - Workflow apps: `8`
 - Docs: `58`
-- Smoke tests: `236`
+- Smoke tests: `248`
 - Repository: https://github.com/coo1white/cool-workflow
 
 ## Architecture
@@ -125,6 +125,7 @@ multi-agent host -> topology -> blackboard/coordinator
 - [core/util/numeric-flag.ts](../src/core/util/numeric-flag.ts)
 - [mcp/dispatch.ts](../src/mcp/dispatch.ts)
 - [mcp/server.ts](../src/mcp/server.ts)
+- [mcp/tool-process.ts](../src/mcp/tool-process.ts)
 - [shell/agent-config.ts](../src/shell/agent-config.ts)
 - [shell/app-run-cli.ts](../src/shell/app-run-cli.ts)
 - [shell/audit-cli.ts](../src/shell/audit-cli.ts)
@@ -166,6 +167,7 @@ multi-agent host -> topology -> blackboard/coordinator
 - [shell/observability.ts](../src/shell/observability.ts)
 - [shell/onramp.ts](../src/shell/onramp.ts)
 - [shell/operator-ux-text.ts](../src/shell/operator-ux-text.ts)
+- [shell/perf-trace.ts](../src/shell/perf-trace.ts)
 - [shell/pipeline-cli.ts](../src/shell/pipeline-cli.ts)
 - [shell/pipeline.ts](../src/shell/pipeline.ts)
 - [shell/reclamation-io.ts](../src/shell/reclamation-io.ts)
@@ -282,6 +284,7 @@ multi-agent host -> topology -> blackboard/coordinator
 
 Smoke tests mirror the public contracts. The high-signal suites are:
 
+- [agent-auto-detect-gemini-native-smoke.js](../test/agent-auto-detect-gemini-native-smoke.js)
 - [agent-backend-concurrent-user-env-smoke.js](../test/agent-backend-concurrent-user-env-smoke.js)
 - [agent-backend-sandbox-deny-smoke.js](../test/agent-backend-sandbox-deny-smoke.js)
 - [agent-backend-user-env-smoke.js](../test/agent-backend-user-env-smoke.js)
@@ -382,13 +385,18 @@ Smoke tests mirror the public contracts. The high-signal suites are:
 - [ledger-verify-smoke.js](../test/ledger-verify-smoke.js)
 - [loop-bounded-expansion-smoke.js](../test/loop-bounded-expansion-smoke.js)
 - [man-run-registry-traversal-smoke.js](../test/man-run-registry-traversal-smoke.js)
+- [manifest-vendor-registry-failclosed-smoke.js](../test/manifest-vendor-registry-failclosed-smoke.js)
 - [mcp-app-surface-smoke.js](../test/mcp-app-surface-smoke.js)
+- [mcp-control-process-responsiveness-smoke.js](../test/mcp-control-process-responsiveness-smoke.js)
 - [mcp-ping-and-arg-coercion-smoke.js](../test/mcp-ping-and-arg-coercion-smoke.js)
 - [mcp-protocol-version-smoke.js](../test/mcp-protocol-version-smoke.js)
 - [mcp-queue-epipe-smoke.js](../test/mcp-queue-epipe-smoke.js)
 - [mcp-surface-registry-smoke.js](../test/mcp-surface-registry-smoke.js)
+- [mcp-tool-authority-policy-smoke.js](../test/mcp-tool-authority-policy-smoke.js)
 - [mcp-tool-call-coverage-smoke.js](../test/mcp-tool-call-coverage-smoke.js)
 - [mcp-tool-call-error-isresult-smoke.js](../test/mcp-tool-call-error-isresult-smoke.js)
+- [mcp-tool-process-lifecycle-smoke.js](../test/mcp-tool-process-lifecycle-smoke.js)
+- [mcp-tool-shutdown-containment-smoke.js](../test/mcp-tool-shutdown-containment-smoke.js)
 - [mcp-untrusted-content-advisory-smoke.js](../test/mcp-untrusted-content-advisory-smoke.js)
 - [metrics-summary-limit-smoke.js](../test/metrics-summary-limit-smoke.js)
 - [missing-artifact-gate-live-smoke.js](../test/missing-artifact-gate-live-smoke.js)
@@ -416,6 +424,7 @@ Smoke tests mirror the public contracts. The high-signal suites are:
 - [operator-ux-smoke.js](../test/operator-ux-smoke.js)
 - [orphan-runs-gc-smoke.js](../test/orphan-runs-gc-smoke.js)
 - [parallel-onramp-smoke.js](../test/parallel-onramp-smoke.js)
+- [parity-cli-reachability-smoke.js](../test/parity-cli-reachability-smoke.js)
 - [parity-doc-sync-smoke.js](../test/parity-doc-sync-smoke.js)
 - [parse-guard-smoke.js](../test/parse-guard-smoke.js)
 - [parse-hardening-round2-smoke.js](../test/parse-hardening-round2-smoke.js)
@@ -456,14 +465,17 @@ Smoke tests mirror the public contracts. The high-signal suites are:
 - [robustness-hardening-smoke.js](../test/robustness-hardening-smoke.js)
 - [run-all-agent-env-hermetic-smoke.js](../test/run-all-agent-env-hermetic-smoke.js)
 - [run-all-json-summary-smoke.js](../test/run-all-json-summary-smoke.js)
+- [run-archive-intake-limits-smoke.js](../test/run-archive-intake-limits-smoke.js)
 - [run-export-cross-machine-smoke.js](../test/run-export-cross-machine-smoke.js)
 - [run-export-import-smoke.js](../test/run-export-import-smoke.js)
 - [run-export-restore-rerun-smoke.js](../test/run-export-restore-rerun-smoke.js)
 - [run-export-restore-resume-smoke.js](../test/run-export-restore-resume-smoke.js)
 - [run-fixture-compat-smoke.js](../test/run-fixture-compat-smoke.js)
+- [run-import-canonical-table-smoke.js](../test/run-import-canonical-table-smoke.js)
 - [run-import-path-traversal-smoke.js](../test/run-import-path-traversal-smoke.js)
 - [run-import-tamper-failclosed-smoke.js](../test/run-import-tamper-failclosed-smoke.js)
 - [run-inspect-archive-smoke.js](../test/run-inspect-archive-smoke.js)
+- [run-paths-shell-boundary-smoke.js](../test/run-paths-shell-boundary-smoke.js)
 - [run-registry-control-plane-smoke.js](../test/run-registry-control-plane-smoke.js)
 - [run-restore-failclosed-smoke.js](../test/run-restore-failclosed-smoke.js)
 - [run-resume-drive-smoke.js](../test/run-resume-drive-smoke.js)
@@ -477,6 +489,7 @@ Smoke tests mirror the public contracts. The high-signal suites are:
 - [schedule-routine-daemon-smoke.js](../test/schedule-routine-daemon-smoke.js)
 - [scheduling-routine-lock-concurrency-smoke.js](../test/scheduling-routine-lock-concurrency-smoke.js)
 - [schema-validation-smoke.js](../test/schema-validation-smoke.js)
+- [schema-version-definition-guard-smoke.js](../test/schema-version-definition-guard-smoke.js)
 - [security-trust-hardening-smoke.js](../test/security-trust-hardening-smoke.js)
 - [self-audit-hardening-smoke.js](../test/self-audit-hardening-smoke.js)
 - [sigint-sigterm-drive-loop-smoke.js](../test/sigint-sigterm-drive-loop-smoke.js)
@@ -513,6 +526,7 @@ Smoke tests mirror the public contracts. The high-signal suites are:
 - [verify-import-audit-chain-smoke.js](../test/verify-import-audit-chain-smoke.js)
 - [web-desktop-workbench-smoke.js](../test/web-desktop-workbench-smoke.js)
 - [wiring-lazy-shell-imports-smoke.js](../test/wiring-lazy-shell-imports-smoke.js)
+- [workbench-load-smoke.js](../test/workbench-load-smoke.js)
 - [workbench-port-range-smoke.js](../test/workbench-port-range-smoke.js)
 - [worker-accept-path-architecture-smoke.js](../test/worker-accept-path-architecture-smoke.js)
 - [worker-isolation-smoke.js](../test/worker-isolation-smoke.js)

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

@@ -201,3 +201,5 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.2.4
 
 0.2.5
+
+0.2.6

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

@@ -341,3 +341,5 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.2.4
 
 0.2.5
+
+0.2.6

package/docs/release-tooling.7.md

@@ -169,6 +169,14 @@ LLM step (the reviewer) is **delegated** through CW's agent backend, so whicheve
 model you set up does the review. CW starts the agent argv-style (`shell:false`),
 takes on the agent's own credentials, and pulls in no model SDK — the red line.
 
+The control plane runs the deterministic gate once. It keeps a failed child
+command's short tail on stderr for the operator, but never puts that tail in a
+saved verdict. The reviewer does not run that gate again. It checks judgment
+questions the gate cannot check. A reviewer rejection has exact `REJECTED`,
+`kind: semantic-review`, and numbered facts with repo-relative `file:line`
+places. Other rejection text is invalid reviewer output: the cut stops, no tag
+is made, and it is not a verified code rejection.
+
 ```bash
 # the operator's one-command release (scripts/release-oneclick.js — preflight,
 # gated cut, tag-only push, CI wait, npm confirmation, verdict-record PR):
@@ -322,3 +330,5 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.2.4
 
 0.2.5
+
+0.2.6

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

@@ -123,6 +123,11 @@ adds the drive outcome to the result under a `drive` field. The default (no
 gives `drive.status="blocked"` (fail-closed, never a made-up completion); CW
 hands worker execution to your agent and never runs a model itself.
 
+Human output with `--drive` or `--once` prints the drive status, worker counts,
+each drive step, and the terminal commit id when there is one. This makes a
+blocked or failed drive clear without `--json`. Bare `run resume` keeps its old
+text as it is.
+
 ## Queue
 
 `queue add` adds a durable entry to the end of `$CW_HOME/registry/queue.json` with an
@@ -177,6 +182,15 @@ whose top-level integrity block is *absent* — closing the legacy fail-open sea
 where a stripped-integrity archive imported unverified. Unset (the default) keeps
 legacy integrity-less archives byte-identical; the flag is mechanism, not policy.
 
+An operator can also set `CW_MAX_RUN_ARCHIVE_BYTES`,
+`CW_MAX_RUN_ARCHIVE_FILES`, or `CW_MAX_RUN_ARCHIVE_CONTENT_BYTES`. Each value
+has to be a positive safe integer. The first sets a raw file byte limit before
+the archive is read. The second sets a normalized file row limit. The last sets
+the total decoded content byte limit, checked from declared sizes before decode
+and again while content is decoded. A bad setting or over-limit archive stops
+before a run directory is made. With all three unset, import, inspect, restore,
+and offline bundle check keep their present behavior.
+
 The archive's run id becomes a directory name under `DIR/.cw/runs/`, so import
 also refuses any run id that is not a single safe path segment (`[A-Za-z0-9._-]`,
 with no separator and not the `.` or `..` component) and asserts the resolved run
@@ -216,11 +230,20 @@ before importing a bad archive. It is a true preview of import: under
 turn away) also inspects as `ok:false`; with the env unset (default) an absent integrity
 block is only reported, not failed.
 
+The file table is checked before any write. Every row needs a known role, a
+safe size, a sha256 value, and a relative path. Paths are made portable and
+then have to be unique. `state.json`, `import-manifest.json`, and lock files
+are made by the restore work and are not accepted from an archive. A bad row
+or two rows with the same path make inspect, import, and restore fail closed.
+No run directory is made for that input.
+
 **Restore in one fail-closed step.** `run restore PATH --target DIR [--json]`
-does the whole move-a-run-to-another-machine flow as ONE atomic, fail-closed
-step: it integrity-**inspects** the bundle first (writing nothing), **imports**
-it, then reuses the verification `import` already ran — and reports `ok:true`
-ONLY when that verify passes. This closes a real gap: `run import` runs a
+does the whole move-a-run-to-another-machine flow as one atomic, fail-closed
+step. It integrity-**inspects** the bundle first, imports it into a same-disk
+staging tree, and checks the file, telemetry, trust-audit, and run-state records there.
+Only then does it put final paths into state and publish the run with one
+directory rename. It reports `ok:true` only after a check of the published
+tree. This closes a real gap: `run import` runs a
 verification (it re-proves restored file digests, the **telemetry ledger**, and
 the **trust-audit hash chain**) and reports it, but does NOT fail on it — it
 exits `0` even when that chain does not verify. So a run whose telemetry or
@@ -229,13 +252,14 @@ made-up success. `run restore` refuses exactly that: it fails closed on the same
 verification `import` only reports. A bundle that fails the up-front integrity
 inspect is refused **before any import**, so nothing is written and the run is
 never left part-restored; the result carries `imported:null` and `verify:null`.
-A bundle that imports but fails post-import verification is reported with
-`ok:false` too. It exits `1` whenever `ok:false`, so `cw run restore <path>` is a
+A bundle that imports in staging but fails a chain check is reported with
+`ok:false`, `imported:null`, and no final run. A final run with the same id is
+never merged or replaced. A staging write fault is cleaned up. It exits `1`
+whenever `ok:false`, so `cw run restore <path>` is a
 single command that either lands a fully-proven run or refuses with a non-zero
-exit — never a made-up success. The result is structured
+exit — never a made-up success or a half-written final run. The result is structured
 (`{ schemaVersion, ok, target, inspect, imported, verify, registry }`) so it is
-scriptable. `run import` and `run inspect-archive` are unchanged; restore is a
-thin composition of `inspectArchive` + `importRun` (reusing its verification).
+scriptable. `run import` and `run inspect-archive` are unchanged.
 
 MCP gives the same mechanisms as `cw_run_export`, `cw_run_import`,
 `cw_run_verify_import`, `cw_run_inspect_archive`, and `cw_run_restore`; the CLI
@@ -484,3 +508,5 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.2.4
 
 0.2.5
+
+0.2.6

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

@@ -294,3 +294,5 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.2.4
 
 0.2.5
+
+0.2.6

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

@@ -330,3 +330,5 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.2.4
 
 0.2.5
+
+0.2.6

package/docs/team-collaboration.7.md

@@ -266,3 +266,5 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
 0.2.4
 
 0.2.5
+
+0.2.6

package/docs/trust-audit-anchor.7.md

@@ -73,3 +73,5 @@ security-trust-hardening(7), cli-mcp-parity(7), report-verifiable-bundle(7)
 0.2.4
 
 0.2.5
+
+0.2.6

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

@@ -10,6 +10,28 @@ speed and the MCP server (JSON-RPC over stdio) showed things for machine context
 Workbench lets a person look at a run quickly — and adds NOTHING else. It works out,
 decides, and keeps nothing that the CLI/MCP cannot already make.
 
+## Simple UI rules
+
+The Workbench follows three public sets of rules:
+
+- FreeBSD [style(9)](https://man.freebsd.org/cgi/man.cgi?query=style&sektion=9)
+  says that usage text has to agree with the man page and that error text has
+  to be easy to find. CW keeps command words and error words clear and fixed.
+- FreeBSD [hier(7)](https://man.freebsd.org/cgi/man.cgi?query=hier&sektion=7)
+  gives each sort of file one place. The Workbench keeps no state; `.cw/`
+  stays the one state source.
+- Homebrew [brew(1)](https://docs.brew.sh/Manpage) puts its essential commands
+  before its full command list, and gives `help`, `doctor`, `info`, and
+  `search`. CW gives one clear next command in place of a hidden recovery
+  action. Homebrew [External Commands](https://docs.brew.sh/External-Commands)
+  also shows how a front door may be added without a change to the tool's
+  inside parts. The Workbench stays a view over present CW capabilities.
+- Codex [Best practices](https://learn.chatgpt.com/guides/best-practices.md)
+  puts context, checks, and review in the work path. Its
+  [app commands](https://learn.chatgpt.com/docs/reference/commands.md) keep
+  search, key movement, and Back/Forward movement close at hand. The
+  Workbench keeps the run and panel in the page address for the same reason.
+
 ## The third front door (mechanism vs policy)
 
 The kernel and the durable `.cw/` state are the MECHANISM. The CLI, the MCP
@@ -85,6 +107,7 @@ The host has least privilege and is local by default:
 
 - it binds the loopback interface `127.0.0.1` ONLY — never a public address
 - it is READ-ONLY: every route is `GET`; any write verb is turned away with `405`
+- a read view makes no audit, metrics, or other derived file under `.cw/`
 - it turns away non-localhost `Host` headers (a DNS-rebinding defense) with `403`
 - it turns away path traversal out of `ui/workbench/` with `403`
 - it serves nothing past the current user's `.cw/` scope and the Run Registry's
@@ -120,6 +143,29 @@ cw workbench view <run-id> [--json]         # five-panel WorkbenchRunView for on
 cw workbench serve [--port N] [--scope repo|home] [--once|--json] [--require-token]
 ```
 
+The page address may name one run and one panel:
+
+```text
+#run=<encoded-run-id>&tab=<graph|blackboard|worker|candidate|audit|metrics|collaboration>
+```
+
+The `run` value is needed for a run view. With no `tab`, the page uses
+`graph`. A bad tab name is changed to `graph`. Back, Forward, and page reload
+keep the named view. The Workbench token stays in the page query and is never
+put in this fragment.
+
+The run list and panels work with a keyboard. Left and Right move through the
+panel tabs; Home and End go to the first and last tab. The active tab and its
+panel are joined with the standard ARIA tab and tabpanel links.
+
+When a present panel payload has an `integrity`, `problems`,
+`missingEvidence`, `nextAction`, or `nextActions` fact, the page puts a small
+`What matters` block before the full panel record. It copies only source
+values. It does not make a new state, rank, or action. An empty problem or
+missing-evidence list is shown as `none`. A field with a shape the page does
+not know is left out of the block. The full structured view or JSON stays
+under the block.
+
 `cw workbench serve` with `--once`/`--json` prints the serve descriptor (bind
 host/port, scope, routes) and stops without starting a server; the default starts
 the localhost host (like `schedule daemon`). The MCP tools `cw_workbench_view` and
@@ -304,3 +350,5 @@ _No behavioral change in v0.1.89 (CLI-surface golden-path + help-output fixes on
   next-step line when a run is blocked or failed.
 
 0.2.5
+
+0.2.6

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": "0.2.5",
+    "version": "0.2.6",
     "license": "BSD-2-Clause",
     "homepage": "https://github.com/coo1white/cool-workflow",
     "author": {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "cool-workflow",
-  "version": "0.2.5",
+  "version": "0.2.6",
   "bin": {
     "cool-workflow": "scripts/cw.js",
     "cw": "scripts/cw.js"
@@ -75,7 +75,8 @@
     "test:unit": "node dist/cli.js version > /dev/null && node test/run-unit.js",
     "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",
-    "validate:schema": "node scripts/validate-run-state-schema.js"
+    "validate:schema": "node scripts/validate-run-state-schema.js",
+    "bench:workbench:deep": "node ../../scripts/bench/workbench-load.js"
   },
   "devDependencies": {
     "@types/node": "^26.1.1",

package/scripts/canonical-apps.js

@@ -83,7 +83,7 @@ const canonicalApps = [
       "--source",
       "plugins/cool-workflow/docs/workflow-app-framework.7.md",
       "--scope",
-      "Cool Workflow v0.2.5",
+      "Cool Workflow v0.2.6",
       "--freshness",
       "as of release preparation"
     ]
@@ -117,14 +117,14 @@ function main() {
     assert.ok(summary, `${app.id} must appear in app list`);
     assert.equal(summary.sourceKind, "app-directory");
     assert.equal(summary.legacy, false);
-    assert.equal(summary.version, "0.2.5");
+    assert.equal(summary.version, "0.2.6");
 
     const validation = runJson(["app", "validate", manifestPath]);
     assert.equal(validation.valid, true, `${app.id} manifest must validate`);
 
     const shown = runJson(["app", "show", app.id]);
     assert.equal(shown.app.id, app.id);
-    assert.equal(shown.app.version, "0.2.5");
+    assert.equal(shown.app.version, "0.2.6");
     assert.ok(shown.app.metadata.canonical, `${app.id} must be marked canonical`);
     assert.ok(shown.app.sandboxProfiles.length > 0, `${app.id} must declare sandbox profiles`);
     assertTaskIdsUnique(shown);
@@ -135,7 +135,7 @@ function main() {
     const plan = runJson(["plan", app.id, ...app.args(workspace)]);
     const state = JSON.parse(fs.readFileSync(plan.statePath, "utf8"));
     assert.equal(state.workflow.app.id, app.id);
-    assert.equal(state.workflow.app.version, "0.2.5");
+    assert.equal(state.workflow.app.version, "0.2.6");
     assert.equal(state.workflow.app.metadata.canonical, true);
     assert.ok(state.tasks.some((task) => task.requiresEvidence), `${app.id} plan must include evidence gates`);
     assert.ok(state.tasks.every((task) => task.sandboxProfileId), `${app.id} plan must include sandbox hints`);

package/scripts/dogfood-release.js

@@ -6,7 +6,7 @@ const fs = require("node:fs");
 const path = require("node:path");
 const { CoolWorkflowRunner } = require("../dist/shell/orchestrator.js");
 
-const TARGET_VERSION = "0.2.5";
+const TARGET_VERSION = "0.2.6";
 const PREVIOUS_VERSION = "0.1.31";
 const pluginRoot = path.resolve(__dirname, "..");
 const repoRoot = path.resolve(pluginRoot, "..", "..");