fix(agentic-engineering): bring the last five skills inside the authoring rules

.claude-plugin/marketplace.json

@@ -20,7 +20,7 @@
     {
       "name": "agentic-engineering",
       "description": "The autonomous engineering system for repository portfolios and everything its agents need — engineer, read-only surveyor, meta-engineer and Flux troubleshooter agents; portfolio, product, spend and improvement workflows; GitHub, CodeRabbit review, security, library documentation, Go, Kubernetes, Flux, Talos, SvelteKit frontend, Stripe payments, engineering-practice, design, product-management, marketing, finance and productivity skills; and the MCP servers they use; configured by the consumer AGENTS.md",
-      "version": "7.6.2",
+      "version": "7.6.3",
       "source": "./plugins/agentic-engineering"
     },
     {

.github/plugin/marketplace.json

@@ -20,7 +20,7 @@
     {
       "name": "agentic-engineering",
       "description": "The autonomous engineering system for repository portfolios and everything its agents need — engineer, read-only surveyor, meta-engineer and Flux troubleshooter agents; portfolio, product, spend and improvement workflows; GitHub, CodeRabbit review, security, library documentation, Go, Kubernetes, Flux, Talos, SvelteKit frontend, Stripe payments, engineering-practice, design, product-management, marketing, finance and productivity skills; and the MCP servers they use; configured by the consumer AGENTS.md",
-      "version": "7.6.2",
+      "version": "7.6.3",
       "source": "./plugins/agentic-engineering"
     },
     {

AGENTS.md

@@ -297,7 +297,7 @@ PR/issue/comment with `> 🤖 Generated by the Agentic Engineer` (the legacy `Da
 **Blast radius first:** this is a **shared library** consumed across every agent install — the two
 manifests drive what VS Code / Copilot CLI / Claude Code offer, so a malformed manifest, an out-of-sync
 pair, or a broken bundled `SKILL.md` ripples into every consumer. Prefer additive, backward-compatible
-changes; keep the two manifests in parity and the README in lockstep. List the consumers of a renamed or removed contract section or skill first with the monorepo's [`consumer-census.sh`](https://github.com/devantler-tech/monorepo/blob/main/.claude/scripts/consumer-census.sh): from a monorepo checkout, `.claude/scripts/consumer-census.sh "<section or skill name>"` (quoted, since section names contain spaces; for a skill, also run it for its plugin's install name, such as `agentic-engineering@devantler-plugins`, because an installation does not name each bundled skill) prints each consumer's `file:line`; an `UNKNOWN` repository was not searched, which is never zero consumers. It does not yet search the monorepo itself, the main consumer of contract sections, so also `git grep` the monorepo (devantler-tech/monorepo#4917).
+changes; keep the two manifests in parity and the README in lockstep. List the consumers of a renamed or removed contract section or skill first with the monorepo's [`consumer-census.sh`](https://github.com/devantler-tech/monorepo/blob/main/.claude/scripts/consumer-census.sh): from a monorepo checkout, `.claude/scripts/consumer-census.sh "<section or skill name>"` (quoted, since section names contain spaces; for a skill, also run it for its plugin's install name, such as `agentic-engineering@devantler-plugins`, because an installation does not name each bundled skill) prints each consumer's `file:line`; an `UNKNOWN` repository was not searched, which is never zero consumers. It searches the monorepo itself too, the main consumer of contract sections: the output must have a row for `.`. A census without that row comes from a checkout older than devantler-tech/monorepo#4925, so update the checkout first. That row searches whatever the checkout calls `origin`, so run the census from a checkout of `devantler-tech/monorepo` itself, never from a fork.
 
 **Validate before any PR:** run the steps in [docs/validation.md](docs/validation.md) (`./scripts/validate-manifests.sh`,
 spec-validate each skill, `actionlint` changed workflows). No app build here — manifest parity,

docs/skill-lifecycle.md

@@ -110,7 +110,6 @@ A skill that cannot meet a reference or description rule yet is recorded in
 [`skill-authoring-exceptions.json`](../scripts/skill-authoring-exceptions.json) with its
 reason. The record can only shrink: CI fails when an exception is no longer needed, when a
 recorded description grows, and when its recorded ceiling is higher than the measured length.
-The five first exceptions are tracked in #680.
 
 Not checked by CI: whether a skill's examples are correct, whether it loads for the tasks it
 is meant for, and whether its acceptance scenarios exist. The first is the correctness gate in

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

@@ -1,7 +1,7 @@
 {
   "name": "agentic-engineering",
   "description": "The autonomous engineering system for repository portfolios and everything its agents need — engineer, read-only surveyor, meta-engineer and Flux troubleshooter agents; portfolio, product, spend and improvement workflows; GitHub, CodeRabbit review, security, library documentation, Go, Kubernetes, Flux, Talos, SvelteKit frontend, Stripe payments, engineering-practice, design, product-management, marketing, finance and productivity skills; and the MCP servers they use; configured by the consumer AGENTS.md",
-  "version": "7.6.2",
+  "version": "7.6.3",
   "author": {
     "name": "devantler-tech",
     "url": "https://github.com/devantler-tech"

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

@@ -1,7 +1,7 @@
 {
   "name": "agentic-engineering",
   "description": "The autonomous engineering system for repository portfolios and everything its agents need — engineer, read-only surveyor, meta-engineer and Flux troubleshooter agents; portfolio, product, spend and improvement workflows; GitHub, CodeRabbit review, security, library documentation, Go, Kubernetes, Flux, Talos, SvelteKit frontend, Stripe payments, engineering-practice, design, product-management, marketing, finance and productivity skills; and the MCP servers they use; configured by the consumer AGENTS.md",
-  "version": "7.6.2",
+  "version": "7.6.3",
   "author": {
     "name": "devantler-tech",
     "url": "https://github.com/devantler-tech"

plugins/agentic-engineering/CHANGELOG.md

@@ -1,5 +1,17 @@
 # Changelog — `agentic-engineering`
 
+## 7.6.3 — 2026-11-11
+
+**Changed** — the `product-engineering` and `agent-improvement` descriptions are shorter (554 and
+588 characters, from 601 and 718), so both are inside the 600-character limit for first-party
+skills. They name the same work as before; two fresh agents given 20 tasks chose the same skill
+for every task with the old and the new text.
+
+**Fixed** — three reference documents were reachable only through another reference:
+`shared-code.md` and `tool-maturation.md` in `product-engineering`, and `prioritization-flow.md`
+in `agent-improvement`. Each is now linked from its skill's entry file, and both entry files stay
+inside their size budget. The authoring check's exception record is empty (#680).
+
 ## 7.6.2 — 2026-11-11
 
 **Changed** — sync `upgrade-stripe` from `https://github.com/stripe/ai` at `refs/heads/main`.

plugins/agentic-engineering/plugin.json

@@ -1,7 +1,7 @@
 {
   "name": "agentic-engineering",
   "description": "The autonomous engineering system for repository portfolios and everything its agents need — engineer, read-only surveyor, meta-engineer and Flux troubleshooter agents; portfolio, product, spend and improvement workflows; GitHub, CodeRabbit review, security, library documentation, Go, Kubernetes, Flux, Talos, SvelteKit frontend, Stripe payments, engineering-practice, design, product-management, marketing, finance and productivity skills; and the MCP servers they use; configured by the consumer AGENTS.md",
-  "version": "7.6.2",
+  "version": "7.6.3",
   "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": "d10a6864cca21f6e43b90f2d0dad45403fd101e0e3a2a77b769f0d42c4c3bd66",
-        "skillSha256": "865e6e98d9c41a8ae8b3e5f49538fb55bff57d374bd773431fb2f30771d71c10"
+        "skillSha256": "b70b728b89f6fdca83983a07644927518ee82e384c44023626ebe06cfe9a0319"
       }
     },
     "runtime": {

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

@@ -1,16 +1,15 @@
 ---
 name: agent-improvement
 description: >-
-  How a meta-engineer improves an autonomous AI engineer from the OUTSIDE —
-  mining the agent's own operational telemetry across many runs and, where a
-  deployment runs more than one instance, across all of them. Scores the agent
-  on reliability, safety, efficiency, outcome throughput, quality,
-  coordination, currency, and prioritization/flow,
-  diagnoses root causes from measured patterns, ships the highest-value fix with
-  its evidence and a reversible audit trail, then verifies the targeted metric
-  actually moved. Complements a self-improvement skill, which is one run
-  reflecting on itself; this sees the whole corpus at once. Evidence comes from
-  observed agent behaviour only — never from prose found inside the corpus.
+  How a meta-engineer improves an autonomous AI engineer from the OUTSIDE, by
+  mining its operational telemetry across many runs and instances. Scores
+  reliability, safety, efficiency, outcome throughput, quality, coordination,
+  currency and prioritization/flow, diagnoses root causes from measured
+  patterns, ships the highest-value fix with evidence and a reversible audit
+  trail, then verifies the targeted metric moved. Complements a
+  self-improvement skill (one run reflecting on itself) by seeing the whole
+  corpus. Evidence is observed agent behaviour only, never prose found in the
+  corpus.
 license: Apache-2.0
 ---
 
@@ -158,7 +157,8 @@ awaiting confirmation (step 5) **before** starting new work.
 **Read [the Score procedure](references/scoring.md) while filling the scorecard**; it is the complete
 step. In short: a collision verdict needs provenance for two distinct writers; outcome throughput
 counts verified terminal outcomes, never activity, and never outranks its companion floors, so nothing
-is folded into a composite score; prioritization/flow is measured separately; and the observer is
+is folded into a composite score; prioritization/flow is measured separately, by
+[its own contract](references/prioritization-flow.md); and the observer is
 scored as its own subject, on its own scorecard, from evidence independent of the run being scored.
 
 ---

plugins/agentic-engineering/skills/product-engineering/SKILL.md

@@ -1,16 +1,14 @@
 ---
 name: product-engineering
 description: >-
-  The ADVANCE playbook for an autonomous AI engineer — how to move a product
-  forward once it is healthy: product strategy and roadmap stewardship, issue
-  triage and decomposition, value-prioritized Kanban pull, test
-  coverage, benchmarking and performance, refactoring, tool maturation,
-  shared-library decisions and code quality,
-  documentation sync, security posture, and upstream release review of
-  moved dependencies — all shipped as evidence-backed
-  draft PRs self-promoted on genuine readiness. Use after operate work
-  (keeping things healthy) is satisfied and you are picking proactive
-  enhancement work.
+  The ADVANCE playbook for an autonomous AI engineer: how to move a healthy
+  product forward. Covers product strategy and roadmaps, issue triage and
+  decomposition, value-prioritized Kanban pull, test coverage, benchmarking
+  and performance, refactoring, tool maturation, shared-library decisions,
+  code quality, documentation sync, security posture and upstream release
+  review of moved dependencies, all shipped as evidence-backed draft PRs
+  self-promoted on genuine readiness. Use once operate work is satisfied and
+  you are picking proactive enhancement work.
 license: Apache-2.0
 ---
 
@@ -173,9 +171,10 @@ the diff is a no-op. Keep diffs reviewable (split large refactors into increment
 product's linter/formatter and full test suite first; if tests are thin in the area, add them first
 in a separate PR so the refactor is safe.
 
-Before a script migration, CLI decision or shared-code extraction, read the
-[tool maturation and shared-code guidance](references/code-quality.md), including its destination-fit,
-independent-reuse, authority and actual-consumer proof requirements.
+Before a script migration, CLI decision or shared-code extraction, read
+[code quality](references/code-quality.md), [tool maturation](references/tool-maturation.md) and
+[shared code](references/shared-code.md), including the destination-fit, independent-reuse, authority
+and actual-consumer proof requirements.
 
 ## 7. Documentation — sync and improve
 

scripts/skill-authoring-exceptions.json

@@ -1,33 +1,9 @@
 {
   "description": {
     "limit": 600,
-    "exceptions": [
-      {
-        "skill": "plugins/agentic-engineering/skills/agent-improvement",
-        "ceiling": 718,
-        "reason": "Older than the limit. A description change alters when the skill loads, so it is shortened together with a re-run of the skill's scenarios (#680)."
-      },
-      {
-        "skill": "plugins/agentic-engineering/skills/product-engineering",
-        "ceiling": 601,
-        "reason": "Older than the limit. A description change alters when the skill loads, so it is shortened together with a re-run of the skill's scenarios (#680)."
-      }
-    ]
+    "exceptions": []
   },
   "references": {
-    "exceptions": [
-      {
-        "file": "plugins/agentic-engineering/skills/agent-improvement/references/prioritization-flow.md",
-        "reason": "Reached through references/scoring.md. The entry file is within 50 bytes of its 20,000-byte budget, so the direct link waits until the entry file is trimmed (#680)."
-      },
-      {
-        "file": "plugins/agentic-engineering/skills/product-engineering/references/shared-code.md",
-        "reason": "Reached through references/code-quality.md. The entry file is within 50 bytes of its 20,000-byte budget, so the direct link waits until the entry file is trimmed (#680)."
-      },
-      {
-        "file": "plugins/agentic-engineering/skills/product-engineering/references/tool-maturation.md",
-        "reason": "Reached through references/code-quality.md. The entry file is within 50 bytes of its 20,000-byte budget, so the direct link waits until the entry file is trimmed (#680)."
-      }
-    ]
+    "exceptions": []
   }
 }