chore(deps): update agent skills (plugins/agentic-engineering/skills/agent-improvement)

.claude-plugin/marketplace.json

@@ -27,7 +27,7 @@
     {
       "name": "agentic-engineering",
       "description": "The autonomous engineering system for repository portfolios — engineer, read-only surveyor, and meta-engineer agents; portfolio, product, spend, and improvement workflows; cross-tool instruction architecture and skill discovery; configured by the consumer AGENTS.md",
-      "version": "6.1.10",
+      "version": "6.1.11",
       "source": "./plugins/agentic-engineering"
     },
     {

.github/plugin/marketplace.json

@@ -27,7 +27,7 @@
     {
       "name": "agentic-engineering",
       "description": "The autonomous engineering system for repository portfolios — engineer, read-only surveyor, and meta-engineer agents; portfolio, product, spend, and improvement workflows; cross-tool instruction architecture and skill discovery; configured by the consumer AGENTS.md",
-      "version": "6.1.10",
+      "version": "6.1.11",
       "source": "./plugins/agentic-engineering"
     },
     {

plugins/agentic-engineering/.claude-plugin/plugin.json

@@ -1,7 +1,7 @@
 {
   "name": "agentic-engineering",
   "description": "The autonomous engineering system for repository portfolios — engineer, read-only surveyor, and meta-engineer agents; portfolio, product, spend, and improvement workflows; cross-tool instruction architecture and skill discovery; configured by the consumer AGENTS.md",
-  "version": "6.1.10",
+  "version": "6.1.11",
   "author": {
     "name": "devantler-tech",
     "url": "https://github.com/devantler-tech"

plugins/agentic-engineering/CHANGELOG.md

@@ -21,6 +21,10 @@ The format follows [Keep a Changelog](https://keepachangelog.com/en/1.1.0/) and
 new one before re-enabling unattended writes. Stopping early resumes writes with the retired FinOps
 schedule still armed, or with a schedule pointing at an entrypoint that no longer resolves.
 
+## 6.1.11 — 2026-10-05
+
+**Changed** — sync `agent-improvement` from `https://github.com/devantler-tech/agent-skills` at `refs/tags/v1.21.29`.
+
 ## 6.1.10 — 2026-10-05
 
 **Fixed** — read-only command admission rejects unresolved quoted special parameters, inspects filter programs after option terminators, and requires Git suppression flags to be actual options. Runtime JSON observations reject unpaired Unicode surrogate escapes before comparing identities.

plugins/agentic-engineering/plugin.json

@@ -1,7 +1,7 @@
 {
   "name": "agentic-engineering",
   "description": "The autonomous engineering system for repository portfolios — engineer, read-only surveyor, and meta-engineer agents; portfolio, product, spend, and improvement workflows; cross-tool instruction architecture and skill discovery; configured by the consumer AGENTS.md",
-  "version": "6.1.10",
+  "version": "6.1.11",
   "author": {
     "name": "devantler-tech",
     "url": "https://github.com/devantler-tech"

plugins/agentic-engineering/resources/provider-neutral.desired-state.json

@@ -82,7 +82,7 @@
         "enabledWhen": "Both optional consumer contract sections are present",
         "mode": "separate-schedule-or-on-demand",
         "definitionSha256": "5bf3fb5e6e99e82900f23b5f42d41bc26a4ea764862a1ca155557162196c26b9",
-        "skillSha256": "1be5fb94932d01f351b49f928b24fcd917864cb8f601112123bdd002954bc953"
+        "skillSha256": "19eeac53f2f8979e8af77f8a5812bf6aa84bd54f3bf457ae2758989fecde106d"
       }
     },
     "runtime": {

plugins/agentic-engineering/skills/agent-improvement/SKILL.md

@@ -3,9 +3,9 @@ description: How a meta-engineer improves an autonomous AI engineer from the OUT
 license: Apache-2.0
 metadata:
     github-path: agent-improvement
-    github-ref: refs/tags/v1.21.27
+    github-ref: refs/tags/v1.21.29
     github-repo: https://github.com/devantler-tech/agent-skills
-    github-tree-sha: 22190a437a20b98dae315841a4a2f569f27f8a57
+    github-tree-sha: c5d8b7638564bcd58eed8f1c40aff283151b7e91
 name: agent-improvement
 ---
 # Agent-improvement loop

plugins/agentic-engineering/skills/agent-improvement/scripts/measure-flow.jq

@@ -99,7 +99,7 @@ def selection_metric:
        {state: "UNKNOWN", oldestUnstarted: null}
      elif .selectedClass != "easy" then
        {state: "NOT-APPLICABLE", oldestUnstarted: null}
-     elif (.candidatesComplete | not) or any(.candidates[];
+     elif (.candidatesComplete | not) or any(.candidates[] | select(.id != $selection.selectedId);
          .actionable == null or (.actionable == true and .startedByEnd == null)) then
        {state: "UNKNOWN", oldestUnstarted: null}
      else