@open-agent-toolkit/cli 0.2.68 → 0.2.69

package/assets/agents/oat-reviewer.md

@@ -1,6 +1,6 @@
 ---
 name: oat-reviewer
-version: 1.2.3
+version: 1.2.4
 description: Unified reviewer for OAT projects - mode-aware verification of requirements/design alignment and code quality. Writes a review artifact to disk by default, or returns structured findings in-memory when dispatched in structured-output mode.
 tools: Read, Bash, Grep, Glob, Write, Task
 color: yellow
@@ -235,6 +235,25 @@ Treat the artifact as a product deliverable. Verify it is:
    - Coverage of design/discovery: every in-scope design component or discovery decision is mapped to at least one task or explicitly deferred/out of scope.
    - Parallelism-claim sanity: any parallel phase group or parallelism statement is consistent with declared file boundaries and dependency order.
 
+6. **Necessary**
+   - Evaluate whether the proposed approach materially serves the requested
+     outcome. For substantial added machinery — a script, schema, persisted
+     record, gate, provider seam, or new document that someone will have to
+     own — identify its consumer (an agent acting on a named instruction, a
+     human reading a named surface, or code at a named call site) and consider
+     whether a smaller approach preserves the requirements.
+   - Judge prose by the behavior it meaningfully guides. An instruction whose
+     consumer is the agent that reads it is a real consumer; missing automated
+     enforcement of that instruction is not proof of uselessness.
+   - A mechanism whose only consumer is its own tests, its own plan, or a
+     deferred future reader is grounds to investigate, not an automatic
+     severity: grade the finding by wasted effort, maintenance burden, or
+     failure to meet the requested outcome, and say which.
+   - Report justified concerns in the existing findings sections. Do not
+     produce an inventory of unproblematic mechanisms, and do not override an
+     explicitly stated user requirement with a preference for simplicity;
+     name the conflict instead.
+
 ### Step 5: Verify Design Alignment
 
 This step applies to **code reviews** only.

package/assets/bundle-metadata.json

@@ -1,4 +1,4 @@
 {
   "schemaVersion": 1,
-  "oatVersion": "0.2.68"
+  "oatVersion": "0.2.69"
 }

package/assets/public-package-versions.json

@@ -1,6 +1,6 @@
 {
-  "cli": "0.2.68",
-  "docs-config": "0.2.68",
-  "docs-theme": "0.2.68",
-  "docs-transforms": "0.2.68"
+  "cli": "0.2.69",
+  "docs-config": "0.2.69",
+  "docs-theme": "0.2.69",
+  "docs-transforms": "0.2.69"
 }

package/assets/skills/oat-wave-execute/SKILL.md

@@ -6,7 +6,7 @@ disable-model-invocation: false
 user-invocable: true
 allowed-tools: Read, Write, Edit, Bash, Grep, Glob, Task
 metadata:
-  version: 1.9.2
+  version: 1.9.3
 ---
 
 # Execute a Wave of External Plans
@@ -235,9 +235,18 @@ writes and stay on the orchestrator's own model class, the same rule
 
 Run the cross-runtime artifact gate with a **bounded** prompt (rule 6): review the
 wrapper artifacts for plan invariants, contract consistency, frontmatter validity,
-and whether any task restates/narrows its source plan — the external plans are
-immutable inputs, NOT review targets. Disposition findings in-artifact
-(gate-invoked artifact review) and commit. A plan gate MAY PROCEED at
+and whether any task restates/narrows its source plan. The external plans'
+content is immutable to the wave (the gate never rewrites a plan), but a plan is
+a legitimate target for one question: whether its proposed mechanism materially
+serves the requested outcome. A reviewer who finds substantial machinery with no
+identifiable consumer — an agent acting on a named instruction, a human reading
+a named surface, or code at a named call site — or a smaller approach that
+preserves the requirements, reports it as a necessity finding. The root returns
+that plan to its author (`oat-repo-improve`) for disposition and defers it from
+this wave until dispositioned; it never silently drops or narrows it, and an
+explicitly stated user requirement is honored over a preference for simplicity.
+Disposition the remaining findings in-artifact (gate-invoked artifact review)
+and commit. A plan gate MAY PROCEED at
 `fixes_completed` per the wave-0/1 precedent, but that is a proceed point, not a
 terminal state. Every gate row MUST flip to `passed` once all fix dispositions
 carry the stored verification records required by the fix-disposition contract below;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@open-agent-toolkit/cli",
-  "version": "0.2.68",
+  "version": "0.2.69",
   "private": false,
   "description": "Open Agent Toolkit CLI",
   "homepage": "https://github.com/voxmedia/open-agent-toolkit/tree/main/packages/cli",
@@ -34,7 +34,7 @@
     "ora": "^9.0.0",
     "yaml": "2.8.2",
     "zod": "^3.25.76",
-    "@open-agent-toolkit/control-plane": "0.2.68"
+    "@open-agent-toolkit/control-plane": "0.2.69"
   },
   "devDependencies": {
     "@types/node": "^22.10.0",