@mrciphersmith/keryx 0.2.71 → 0.2.72

package/src/gdskills/bundled/skills/review/code-learned-review/SKILL.zed.md

@@ -0,0 +1,252 @@
+---
+name: code-learned-review
+description: "Reviews the current branch against conventions this project learned from its own pull-request comments, recorded in a local project skill. Ships with an empty checklist: the content comes from `keryx review learn`, never from the tool. Use when: a project has a learned review skill and wants its own accumulated conventions applied."
+triggers:
+  - "learned review"
+  - "review with our conventions"
+  - "review using what we learned"
+metadata:
+  author: "keryx"
+  version: "2.0.0"
+  category: "review"
+  compatible_harnesses: "cursor,codex,zed,opencode"
+license: "MIT"
+---
+
+# Learned Code Review (current branch only)
+
+## What this skill is, and what it is not
+
+This skill carries **no review conventions of its own**. It is a mechanism with
+an empty checklist, and the checklist is filled in by the project that installs
+it.
+
+What it applies is a **project skill** under
+`.metaproject/project-skills/<module>/<skill>/SKILL.md`, whose `Review Lessons`,
+`Review Checklist` and `Anti-patterns` sections were written from pull-request
+comments left by the people that project named. Point a second project at
+different reviewers and its learned skill diverges from the first. That is the
+intended behaviour: there is no single correct checklist, and shipping one would
+mean shipping one team's opinions to everybody.
+
+If this project has no learned skill yet, **say so and stop**. An empty learned
+review is not a generic review; use `review-orchestrator` for that.
+
+## Workflow
+
+Copy this checklist and track progress:
+
+```
+Learned Review Progress:
+- [ ] Step 1: Read .metaproject/review-learning.config.json — which skill is this project's?
+- [ ] Step 2: Read that project skill; if it has no lessons, stop and say so
+- [ ] Step 3: Determine parent branch and calculate merge-base
+- [ ] Step 4: Collect git diff (committed + local changes)
+- [ ] Step 5: Apply the learned checklist, lesson by lesson
+- [ ] Step 6: Report each finding with the learned lesson it rests on
+- [ ] Step 7: Feed new pull-request comments back with `keryx review learn`
+```
+
+## Mandatory rules
+
+1. **Default scope (no commit hash/range given)**: the review covers **all**
+   changes on the current branch from the merge-base with its parent branch —
+   committed (`BASE_SHA..HEAD`) and local uncommitted (staged/unstaged/untracked).
+2. **Explicit commit hash/range**: review only the requested range; do not add
+   local uncommitted changes unless asked separately.
+3. **The checklist is the project's, not this file's**: every finding cites a
+   line in the project skill named by
+   `.metaproject/review-learning.config.json`. A finding with nothing behind it
+   there is a generic finding and belongs to the reviewer that owns it —
+   `review-logic`, `review-architecture`, `review-security-code`.
+4. **Result**: a detailed report with an explanation per finding and a concrete
+   fix (a minimal patch where the fix is obvious).
+
+## Where the checklist comes from
+
+```bash
+# 1. Which local skill does this project teach, from which repository,
+#    and whose comments count?
+cat .metaproject/review-learning.config.json
+
+# 2. The comments themselves, collected once per round and kept on disk.
+keryx review comments collect --repo <owner/repo> --pr <n> --sha <head-sha>
+
+# 3. The join: configured authors only, read from the record above, never
+#    re-fetched. Writes a proposal and changes nothing.
+keryx review learn --pr <n>
+
+# 4. The only writer. Bumps the skill version, appends to skill-changelog.md,
+#    and refuses any target outside .metaproject/project-skills/.
+keryx skills learn apply .metaproject/data/gdskills/proposals/<id>.json
+```
+
+An author the config does not name contributes nothing — their text reaches no
+proposal and no `SKILL.md`. A project with no config file does not learn, and
+that is a supported state rather than an error.
+
+## Determining the parent ref (deterministic — the branch parent, not the upstream)
+
+Use the first variant that exists:
+
+1. `origin/main`
+2. `origin/master`
+3. `main`
+4. `master`
+5. `@{upstream}` **only when it is not the current feature branch**
+
+```bash
+BRANCH="$(git rev-parse --abbrev-ref HEAD)"
+UPSTREAM_REF="$(git rev-parse --abbrev-ref --symbolic-full-name @{upstream} 2>/dev/null || true)"
+
+PARENT=""
+if git rev-parse --verify -q "origin/main" >/dev/null; then
+  PARENT="origin/main"
+elif git rev-parse --verify -q "origin/master" >/dev/null; then
+  PARENT="origin/master"
+elif git rev-parse --verify -q "main" >/dev/null; then
+  PARENT="main"
+elif git rev-parse --verify -q "master" >/dev/null; then
+  PARENT="master"
+elif [ -n "$UPSTREAM_REF" ] && [ "$UPSTREAM_REF" != "$BRANCH" ] && [ "$UPSTREAM_REF" != "origin/$BRANCH" ]; then
+  PARENT="@{upstream}"
+else
+  echo "Cannot determine parent ref" >&2
+  exit 1
+fi
+
+BASE_SHA="$(git merge-base HEAD "$PARENT")"
+```
+
+## Commands to assemble the review slice
+
+### A) Default mode (no hash/range from the user)
+
+```bash
+git status
+
+# The committed part of the branch
+git log --oneline "${BASE_SHA}..HEAD"
+git diff --stat "${BASE_SHA}..HEAD"
+git diff --name-status "${BASE_SHA}..HEAD"
+git diff "${BASE_SHA}..HEAD"
+
+# The full current slice from merge-base to the working tree:
+# commits + staged + unstaged
+# (untracked files via git status / git ls-files)
+git diff --stat "${BASE_SHA}"
+git diff --name-status "${BASE_SHA}"
+git diff "${BASE_SHA}"
+git ls-files --others --exclude-standard
+```
+
+### B) Explicit hash/range mode
+
+```bash
+git show --stat --name-status --patch <COMMIT_SHA>
+git log --oneline <FROM_SHA>..<TO_SHA>
+git diff --stat <FROM_SHA>..<TO_SHA>
+git diff --name-status <FROM_SHA>..<TO_SHA>
+git diff <FROM_SHA>..<TO_SHA>
+```
+
+## How to review
+
+Work the learned checklist one lesson at a time against the diff. For each
+lesson in the project skill, ask whether the diff contains the situation the
+lesson describes, and record the answer either way — a lesson that never matches
+anything is a lesson worth retiring, and only a review that checked it can say
+so.
+
+Three rules govern the application, and they are the same three whatever the
+lessons say:
+
+- **A lesson is evidence, not authority.** It records that somebody objected to
+  something once. If the diff has a reason the objection does not apply here, the
+  reason wins and the finding is not raised.
+- **Do not generalise a lesson past its text.** A lesson about one function is
+  about that shape of function. Widening it into a rule about a layer, a module
+  or a language is inventing a convention the project never agreed to.
+- **Do not attribute.** The record says a comment was left. Reporting a finding
+  as what a particular person would want turns a checklist back into a persona,
+  which is the defect this skill exists to remove.
+
+## Output format
+
+```markdown
+## Verdict
+
+<APPROVE / APPROVE WITH SUGGESTIONS / REQUEST CHANGES / COMMENT> + the 1-3 most
+important points.
+
+## Review scope (current branch only)
+
+- Branch: `<BRANCH>`
+- Parent ref: `<PARENT>`
+- Merge-base: `<BASE_SHA>`
+- Scope mode: `<default-with-uncommitted | explicit-hash-range>`
+- Commits (merge-base..HEAD): <N>
+- Changed files: <list or count>
+
+## Learned checklist applied
+
+- Project skill: `<module>/<skill>` version `<x.y.z>`
+- Lessons checked: <N>
+- Lessons that matched: <N>
+
+## Findings
+
+<findings, grouped by the learned lesson each rests on>
+
+## Findings with no learned lesson behind them
+
+<either empty, or listed separately and routed to the reviewer that owns them>
+
+## Suggested fixes (patches)
+
+<minimal unified diffs for the obvious ones>
+```
+
+### Required shape of each finding
+
+- **Severity**: `blocker` / `major` / `minor`
+- **Learned lesson**: the line from the project skill this rests on
+- **Location**: file path + the relevant hunk from the diff
+- **Problem**: what is wrong
+- **Why it matters**: correctness / testability / conventions / maintainability
+- **Suggested fix**: what specifically to do, without widening scope
+- **Optional patch**: a unified diff when the fix is simple
+
+Example patch block:
+
+```diff
+diff --git a/path/file.ts b/path/file.ts
+index 0000000..1111111 100644
+--- a/path/file.ts
++++ b/path/file.ts
+@@ -1,3 +1,3 @@
+-const x = 1;
++const x = 2;
+```
+
+## Job Context Awareness
+
+When dispatched by `job-orchestrator` as part of a job pipeline, the prompt MAY
+include:
+
+```
+JOB_NAME:     <job-name>
+CONTEXT_PATH: .metaproject/jobs/<job-name>/ai/context.md
+```
+
+If provided and the file exists, read the context document before starting the
+review. Use it to:
+
+- Understand which libraries and patterns were intentionally chosen for the
+  implementation
+- Avoid flagging correct library usage as issues
+- Provide more accurate findings by understanding the project's architectural
+  decisions
+
+If the file does not exist or is not provided, proceed normally — context is
+optional and non-blocking.

package/src/gdskills/bundled/skills/review/code-mobx-store-review/SKILL.md

@@ -237,7 +237,7 @@ This skill covers **MobX store and state logic** — targeted review of store in
 | Store structure, actions, computed, reactions, async runInAction | ✅ YES | — |
 | View↔Store boundary violations | ✅ YES | — |
 | General code quality, readability, tests | ❌ NO | `code-ai-review` |
-| b091-style logic enforcement | ❌ NO | `code-b091-review` |
+| Learned-convention enforcement | ❌ NO | `code-learned-review` |
 | Naming/style/architecture patterns outside stores | ❌ NO | `code-style-review` |
 
 ---

package/src/gdskills/bundled/skills/review/code-style-review/SKILL.md

@@ -146,7 +146,7 @@ This skill covers **code style and architecture patterns** — naming, organizat
 | Naming conventions, file organization, import order | ✅ YES | — |
 | TypeScript patterns, component/hook structure | ✅ YES | — |
 | Architecture pattern compliance | ✅ YES | — |
-| Logic correctness, type safety deep dives | ❌ NO | `code-ai-review` or `code-b091-review` |
+| Logic correctness, type safety deep dives | ❌ NO | `code-ai-review` or `code-learned-review` |
 | MobX store internals | ❌ NO | `code-mobx-store-review` |
 
 ---

package/src/gdskills/bundled/skills/review/review-architecture/SKILL.md

@@ -211,7 +211,7 @@ Focus on **SRP** and **DIP** — the two most commonly violated in practice.
 
 ## Orchestrated Review Contract
 
-When dispatched by `review-orchestrator`, follow the provided `reviewer-input.schema.json` payload. Return a `REVIEW_RESULT` object compatible with `skills/review-orchestrator/reviewer-finding.schema.json`, then a concise markdown summary. Keep findings evidence-based, include concrete `suggested_fix` for every blocker/major, and return `NEEDS_CONTEXT` instead of guessing when required context is missing.
+When dispatched by `review-orchestrator`, follow the provided `reviewer-input.schema.json` payload. Return a `REVIEW_RESULT` object compatible with `skills/review/review-orchestrator/reviewer-finding.schema.json`, then a concise markdown summary. Keep findings evidence-based, include concrete `suggested_fix` for every blocker/major, and return `NEEDS_CONTEXT` instead of guessing when required context is missing.
 
 ---
 

package/src/gdskills/bundled/skills/review/review-backend/SKILL.md

@@ -230,7 +230,7 @@ is, and a condition that cannot name one is not a `blocker`.
 
 ## Orchestrated Review Contract
 
-When dispatched by `review-orchestrator`, follow the provided `reviewer-input.schema.json` payload. Return a `REVIEW_RESULT` object compatible with `skills/review-orchestrator/reviewer-finding.schema.json`, then a concise markdown summary. Keep findings evidence-based, include concrete `suggested_fix` for every blocker/major, and return `NEEDS_CONTEXT` instead of guessing when required context is missing.
+When dispatched by `review-orchestrator`, follow the provided `reviewer-input.schema.json` payload. Return a `REVIEW_RESULT` object compatible with `skills/review/review-orchestrator/reviewer-finding.schema.json`, then a concise markdown summary. Keep findings evidence-based, include concrete `suggested_fix` for every blocker/major, and return `NEEDS_CONTEXT` instead of guessing when required context is missing.
 
 ---
 

package/src/gdskills/bundled/skills/review/review-clean-code/SKILL.md

@@ -367,7 +367,7 @@ Flags:
 
 ## Orchestrated Review Contract
 
-When dispatched by `review-orchestrator`, follow the provided `reviewer-input.schema.json` payload. Return a `REVIEW_RESULT` object compatible with `skills/review-orchestrator/reviewer-finding.schema.json`, then a concise markdown summary. Keep findings evidence-based, include concrete `suggested_fix` for every blocker/major, and return `NEEDS_CONTEXT` instead of guessing when required context is missing.
+When dispatched by `review-orchestrator`, follow the provided `reviewer-input.schema.json` payload. Return a `REVIEW_RESULT` object compatible with `skills/review/review-orchestrator/reviewer-finding.schema.json`, then a concise markdown summary. Keep findings evidence-based, include concrete `suggested_fix` for every blocker/major, and return `NEEDS_CONTEXT` instead of guessing when required context is missing.
 
 ---
 

package/src/gdskills/bundled/skills/review/review-core-boundaries/SKILL.md

@@ -69,7 +69,7 @@ boundary is the trigger-and-outcome test.
 
 ## Orchestrated Review Contract
 
-When dispatched by `review-orchestrator`, follow the provided `reviewer-input.schema.json` payload. Return a `REVIEW_RESULT` object compatible with `skills/review-orchestrator/reviewer-finding.schema.json`, then a concise markdown summary. Keep findings evidence-based, include concrete `suggested_fix` for every blocker/major, and return `NEEDS_CONTEXT` instead of guessing when required context is missing.
+When dispatched by `review-orchestrator`, follow the provided `reviewer-input.schema.json` payload. Return a `REVIEW_RESULT` object compatible with `skills/review/review-orchestrator/reviewer-finding.schema.json`, then a concise markdown summary. Keep findings evidence-based, include concrete `suggested_fix` for every blocker/major, and return `NEEDS_CONTEXT` instead of guessing when required context is missing.
 
 ---
 

package/src/gdskills/bundled/skills/review/review-flow-graph/SKILL.md

@@ -95,7 +95,7 @@ boundary is the trigger-and-outcome test.
 
 ## Orchestrated Review Contract
 
-When dispatched by `review-orchestrator`, follow the provided `reviewer-input.schema.json` payload. Return a `REVIEW_RESULT` object compatible with `skills/review-orchestrator/reviewer-finding.schema.json`, then a concise markdown summary. Keep findings evidence-based, include concrete `suggested_fix` for every blocker/major, and return `NEEDS_CONTEXT` instead of guessing when required context is missing.
+When dispatched by `review-orchestrator`, follow the provided `reviewer-input.schema.json` payload. Return a `REVIEW_RESULT` object compatible with `skills/review/review-orchestrator/reviewer-finding.schema.json`, then a concise markdown summary. Keep findings evidence-based, include concrete `suggested_fix` for every blocker/major, and return `NEEDS_CONTEXT` instead of guessing when required context is missing.
 
 ---
 

package/src/gdskills/bundled/skills/review/review-frontend/SKILL.md

@@ -630,7 +630,7 @@ If the file does not exist or is not provided, proceed normally — context is o
 
 ## Orchestrated Review Contract
 
-When dispatched by `review-orchestrator`, follow the provided `reviewer-input.schema.json` payload. Return a `REVIEW_RESULT` object compatible with `skills/review-orchestrator/reviewer-finding.schema.json`, then a concise markdown summary. Keep findings evidence-based, include concrete `suggested_fix` for every blocker/major, and return `NEEDS_CONTEXT` instead of guessing when required context is missing.
+When dispatched by `review-orchestrator`, follow the provided `reviewer-input.schema.json` payload. Return a `REVIEW_RESULT` object compatible with `skills/review/review-orchestrator/reviewer-finding.schema.json`, then a concise markdown summary. Keep findings evidence-based, include concrete `suggested_fix` for every blocker/major, and return `NEEDS_CONTEXT` instead of guessing when required context is missing.
 
 ---
 

package/src/gdskills/bundled/skills/review/review-frontend-conventions/SKILL.md

@@ -120,7 +120,7 @@ boundary is the trigger-and-outcome test.
 
 ## Orchestrated Review Contract
 
-When dispatched by `review-orchestrator`, follow the provided `reviewer-input.schema.json` payload. Return a `REVIEW_RESULT` object compatible with `skills/review-orchestrator/reviewer-finding.schema.json`, then a concise markdown summary. Keep findings evidence-based, include concrete `suggested_fix` for every blocker/major, and return `NEEDS_CONTEXT` instead of guessing when required context is missing.
+When dispatched by `review-orchestrator`, follow the provided `reviewer-input.schema.json` payload. Return a `REVIEW_RESULT` object compatible with `skills/review/review-orchestrator/reviewer-finding.schema.json`, then a concise markdown summary. Keep findings evidence-based, include concrete `suggested_fix` for every blocker/major, and return `NEEDS_CONTEXT` instead of guessing when required context is missing.
 
 ---
 

package/src/gdskills/bundled/skills/review/review-highload/SKILL.md

@@ -406,7 +406,7 @@ Flags:
 
 ## Orchestrated Review Contract
 
-When dispatched by `review-orchestrator`, follow the provided `reviewer-input.schema.json` payload. Return a `REVIEW_RESULT` object compatible with `skills/review-orchestrator/reviewer-finding.schema.json`, then a concise markdown summary. Keep findings evidence-based, include concrete `suggested_fix` for every blocker/major, and return `NEEDS_CONTEXT` instead of guessing when required context is missing.
+When dispatched by `review-orchestrator`, follow the provided `reviewer-input.schema.json` payload. Return a `REVIEW_RESULT` object compatible with `skills/review/review-orchestrator/reviewer-finding.schema.json`, then a concise markdown summary. Keep findings evidence-based, include concrete `suggested_fix` for every blocker/major, and return `NEEDS_CONTEXT` instead of guessing when required context is missing.
 
 ---
 

package/src/gdskills/bundled/skills/review/review-logic/SKILL.md

@@ -26,7 +26,7 @@ license: "MIT"
 # Review Logic
 
 Specialized reviewer for **logic correctness and algorithmic soundness**.
-Inherits from the original `code-ai-review` + `code-boss-review` correctness phases
+Inherits from the original `code-ai-review` + `code-learned-review` correctness phases
 and unifies them into a single focused pass.
 
 This reviewer does not care about formatting, naming, or architecture opinions.
@@ -174,7 +174,7 @@ Do not review unrelated files.
 
 ## Orchestrated Review Contract
 
-When dispatched by `review-orchestrator`, follow the provided `reviewer-input.schema.json` payload. Return a `REVIEW_RESULT` object compatible with `skills/review-orchestrator/reviewer-finding.schema.json`, then a concise markdown summary. Keep findings evidence-based, include concrete `suggested_fix` for every blocker/major, and return `NEEDS_CONTEXT` instead of guessing when required context is missing.
+When dispatched by `review-orchestrator`, follow the provided `reviewer-input.schema.json` payload. Return a `REVIEW_RESULT` object compatible with `skills/review/review-orchestrator/reviewer-finding.schema.json`, then a concise markdown summary. Keep findings evidence-based, include concrete `suggested_fix` for every blocker/major, and return `NEEDS_CONTEXT` instead of guessing when required context is missing.
 
 ---
 

package/src/gdskills/bundled/skills/review/review-orchestrator/SKILL.md

@@ -28,7 +28,7 @@ triggers:
   - "review --flow-graph"
   - "review --legacy-profiles"
   - "review --code-ai"
-  - "review --b091"
+  - "review --learned"
   - "review --code-style"
   - "review --mobx-store"
 metadata:
@@ -153,7 +153,7 @@ which of its inputs were recovered and which were never written down.
 
 | Field | Type | Required | Description |
 |-------|------|----------|-------------|
-| `flags` | string[] | no | One or more of: `--frontend`, `--backend`, `--architecture`, `--security`, `--performance`, `--style`, `--clean-code`, `--highload`, `--project-conventions`, `--frontend-conventions`, `--testing-practices`, `--core-boundaries`, `--flow-graph`, `--legacy-profiles`, `--code-ai`, `--b091`, `--code-style`, `--mobx-store`, `--verify`, `--all` |
+| `flags` | string[] | no | One or more of: `--frontend`, `--backend`, `--architecture`, `--security`, `--performance`, `--style`, `--clean-code`, `--highload`, `--project-conventions`, `--frontend-conventions`, `--testing-practices`, `--core-boundaries`, `--flow-graph`, `--legacy-profiles`, `--code-ai`, `--learned`, `--code-style`, `--mobx-store`, `--verify`, `--all` |
 | `path` | string | no | File or directory path to review (e.g., `src/stores/`, `src/components/UserCard.tsx`). Activates **path mode** — reviews the files at this path directly, not a git diff. |
 | `commit_range` | string | no | Explicit commit hash or range (e.g., `abc123..HEAD`). Overrides merge-base detection. Ignored in path mode. |
 | `issue_url` | string | no | GitHub issue or task URL. If provided, Stage 1 gate checks spec compliance before dispatching reviewers. |
@@ -420,7 +420,7 @@ not ask for our reasoning and is reading between other tasks.
 
 ## Review Context Pack
 
-Before routing reviewers, build a compact `review_context` object. This is the shared source of truth for all sub-agents and must follow `skills/review-orchestrator/review-context.schema.json`.
+Before routing reviewers, build a compact `review_context` object. This is the shared source of truth for all sub-agents and must follow `skills/review/review-orchestrator/review-context.schema.json`.
 
 Required content:
 - Request: raw user request, flags, review mode, explicit paths or commit range.
@@ -817,9 +817,9 @@ Legacy/profile reviewers are specialized review profiles that predate the review
 
 | Trigger | Reviewers appended |
 |---|---|
-| `--legacy-profiles` | `code-ai-review` + `code-b091-review` + `code-style-review` + `code-mobx-store-review` when MobX/store files are present |
+| `--legacy-profiles` | `code-ai-review` + `code-learned-review` + `code-style-review` + `code-mobx-store-review` when MobX/store files are present |
 | `--code-ai` | `code-ai-review` |
-| `--b091` | `code-b091-review` |
+| `--learned` | `code-learned-review` |
 | `--code-style` | `code-style-review` |
 | `--mobx-store` | `code-mobx-store-review` |
 | `*.store.ts`, `makeObservable`, `observable`, `computed`, `action.bound` | suggest `code-mobx-store-review` as optional profile reviewer |
@@ -839,13 +839,13 @@ Review Plan Preview must include an `Optional legacy/profile reviewers` group an
 ```text
 Optional legacy/profile reviewers:
   - code-ai-review: available via --code-ai or --legacy-profiles
-  - code-b091-review: available via --b091 or --legacy-profiles
+  - code-learned-review: available via --learned or --legacy-profiles
   - code-style-review: available via --code-style or --legacy-profiles
   - code-mobx-store-review: auto-suggest when *.store.ts or MobX patterns are present; available via --mobx-store or --legacy-profiles
 
 Skipped reviewers:
   - code-ai-review: profile reviewer, not selected unless --code-ai/--legacy-profiles
-  - code-b091-review: profile reviewer, not selected unless --b091/--legacy-profiles
+  - code-learned-review: profile reviewer, not selected unless --learned/--legacy-profiles
   - code-style-review: legacy style profile, not selected unless --code-style/--legacy-profiles
   - code-mobx-store-review: not selected unless --mobx-store/--legacy-profiles or MobX store files are detected
 ```
@@ -989,7 +989,7 @@ Runtime rules:
 
 Do not use vague fallback messages such as "running through available agent types" without naming which reviewers used fallback and why.
 
-Pass each sub-reviewer a payload matching `skills/review-orchestrator/reviewer-input.schema.json`:
+Pass each sub-reviewer a payload matching `skills/review/review-orchestrator/reviewer-input.schema.json`:
 
 ```yaml
 review_context: <bounded context pack>
@@ -1012,7 +1012,7 @@ target_path: <resolved path or file list>
 file_contents: <bounded file contents relevant to this reviewer>
 ```
 
-Each reviewer must return a `REVIEW_RESULT` object matching `skills/review-orchestrator/reviewer-finding.schema.json`, followed by a concise markdown summary. The orchestrator must reject or normalize free-form reports before consolidation.
+Each reviewer must return a `REVIEW_RESULT` object matching `skills/review/review-orchestrator/reviewer-finding.schema.json`, followed by a concise markdown summary. The orchestrator must reject or normalize free-form reports before consolidation.
 
 **Important for path mode:** instruct each reviewer to check the **entire file**, not just changes. The scope report should say "Path: `<TARGET_PATH>`" instead of a branch/merge-base.
 
@@ -1034,7 +1034,7 @@ Each reviewer must return a `REVIEW_RESULT` object matching `skills/review-orche
 | Test / e2e conventions | NO | `review-testing-practices` |
 | Shared core boundary rules | NO | `review-core-boundaries` |
 | Shared flow/graph abstraction contracts | NO | `review-flow-graph` |
-| Legacy/profile review profiles | NO | `code-ai-review`, `code-b091-review`, `code-style-review`, `code-mobx-store-review` |
+| Legacy/profile review profiles | NO | `code-ai-review`, `code-learned-review`, `code-style-review`, `code-mobx-store-review` |
 
 ---
 

package/src/gdskills/bundled/skills/review/review-performance/SKILL.md

@@ -368,7 +368,7 @@ If the file does not exist or is not provided, proceed normally — context is o
 
 ## Orchestrated Review Contract
 
-When dispatched by `review-orchestrator`, follow the provided `reviewer-input.schema.json` payload. Return a `REVIEW_RESULT` object compatible with `skills/review-orchestrator/reviewer-finding.schema.json`, then a concise markdown summary. Keep findings evidence-based, include concrete `suggested_fix` for every blocker/major, and return `NEEDS_CONTEXT` instead of guessing when required context is missing.
+When dispatched by `review-orchestrator`, follow the provided `reviewer-input.schema.json` payload. Return a `REVIEW_RESULT` object compatible with `skills/review/review-orchestrator/reviewer-finding.schema.json`, then a concise markdown summary. Keep findings evidence-based, include concrete `suggested_fix` for every blocker/major, and return `NEEDS_CONTEXT` instead of guessing when required context is missing.
 
 ---
 

package/src/gdskills/bundled/skills/review/review-pr-feedback/SKILL.md

@@ -41,7 +41,7 @@ review-pr-feedback Progress:
 - [ ] Step 4: Group comments by author
 - [ ] Step 5: Classify comment types (blocker intent / suggestion / question / nitpick)
 - [ ] Step 6: Explain each comment + suggest concrete fix with code example
-- [ ] Step 7: Detect senior/boss reviewer comments — offer pattern extraction (with user consent)
+- [ ] Step 7: Comments from configured learning authors — offer a learning proposal (with user consent)
 - [ ] Step 8: Emit structured report with action items checklist
 ```
 
@@ -183,20 +183,39 @@ If confidence is Low, state both interpretations and ask the user which one appl
 
 ---
 
-## Step 7: Senior / Boss Reviewer Handling
-
-**Detection**: if an author is identified as a senior or boss reviewer (by username, or by
-review authority markers such as "REQUEST_CHANGES from `boss`"), trigger this flow:
-
-1. Notify the user: "This PR has comments from a senior reviewer (`<login>`)."
-2. **NEVER update rules, CLAUDE.md, or any configuration without explicit user consent.**
-3. Ask: "Should I analyze `<login>`'s comments for patterns and suggest updates to review rules?"
-4. If the user agrees:
-   - Identify recurring patterns in that reviewer's feedback (not one-off edge cases)
-   - Propose specific additions to `.metaproject/rules/core/code-review-boss-profile.mdc` or the relevant rule file
-   - Present the proposed changes to the user for approval before writing anything
-   - After approval, update the rule file and note it in the output
-5. Extract generalizable patterns only — not personal phrasing or one-off opinions
+## Step 7: Feed the comments back into the project's learned skill
+
+**Detection is configuration, not judgement.** Which authors teach this project is
+declared in `.metaproject/review-learning.config.json`, alongside the local skill
+to teach and the repository to read. Do not decide that an author is senior,
+authoritative, or worth learning from — the file decides, and an author it does
+not name contributes nothing.
+
+If the file is absent, this project does not learn. Say nothing about it and skip
+this step; absence is a supported state, not a misconfiguration.
+
+If the file is present and names at least one author who commented on this PR:
+
+1. Notify the user: "This PR has comments from `<login>`, a configured learning
+   source for `<module>/<skill>`."
+2. **NEVER apply a proposal without explicit user consent.**
+3. Ask: "Should I turn those comments into a learning proposal for
+   `<module>/<skill>`?"
+4. If the user agrees, run the two commands. The first writes a proposal and
+   changes nothing; the second is the only writer:
+   ```bash
+   keryx review learn --pr <n>
+   keryx skills learn apply .metaproject/data/gdskills/proposals/<id>.json
+   ```
+5. Present the proposal's lessons before applying. `learn` reads the record
+   `keryx review comments collect` already wrote and never re-fetches, so the
+   proposal shows exactly what would be written.
+
+**The target is the project skill, never a rule file.** `.metaproject/rules/core/`
+holds shipped templates that `keryx update` overwrites with force, and
+`applyLearningProposal` refuses any target outside
+`.metaproject/project-skills/` — a lesson written anywhere else is lost on the
+next update, or refused outright.
 
 ---
 
@@ -318,7 +337,7 @@ If provided and the file exists, read it before fetching PR comments. If absent,
 
 | Rationalization | Why it is wrong |
 |----------------|-----------------|
-| "I'll update the rule file based on the boss's comments without asking" | NEVER auto-update rules; always ask first — rules affect all future reviews |
+| "I'll write these comments into a rule file without asking" | NEVER apply a learning proposal unsupervised, and never target a rule file — `keryx skills learn apply` refuses anything outside `.metaproject/project-skills/` |
 | "The reviewer's question is just curiosity, not a real concern" | Questions often hide concerns; classify carefully |
 | "I'll skip the 'praise' comments — they're not actionable" | Positive patterns help developers understand what to repeat |
 | "Confidence High for an ambiguous comment" | Low confidence is honest; false confidence leads to wrong fixes |
@@ -327,7 +346,7 @@ If provided and the file exists, read it before fetching PR comments. If absent,
 
 ## Orchestrated Review Contract
 
-When dispatched by `review-orchestrator`, follow the provided `reviewer-input.schema.json` payload. Return a `REVIEW_RESULT` object compatible with `skills/review-orchestrator/reviewer-finding.schema.json`, then a concise markdown summary. Keep findings evidence-based, include concrete `suggested_fix` for every blocker/major, and return `NEEDS_CONTEXT` instead of guessing when required context is missing.
+When dispatched by `review-orchestrator`, follow the provided `reviewer-input.schema.json` payload. Return a `REVIEW_RESULT` object compatible with `skills/review/review-orchestrator/reviewer-finding.schema.json`, then a concise markdown summary. Keep findings evidence-based, include concrete `suggested_fix` for every blocker/major, and return `NEEDS_CONTEXT` instead of guessing when required context is missing.
 
 ---
 

package/src/gdskills/bundled/skills/review/review-security-code/SKILL.md

@@ -373,7 +373,7 @@ If the file does not exist or is not provided, proceed normally — context is o
 
 ## Orchestrated Review Contract
 
-When dispatched by `review-orchestrator`, follow the provided `reviewer-input.schema.json` payload. Return a `REVIEW_RESULT` object compatible with `skills/review-orchestrator/reviewer-finding.schema.json`, then a concise markdown summary. Keep findings evidence-based, include concrete `suggested_fix` for every blocker/major, and return `NEEDS_CONTEXT` instead of guessing when required context is missing.
+When dispatched by `review-orchestrator`, follow the provided `reviewer-input.schema.json` payload. Return a `REVIEW_RESULT` object compatible with `skills/review/review-orchestrator/reviewer-finding.schema.json`, then a concise markdown summary. Keep findings evidence-based, include concrete `suggested_fix` for every blocker/major, and return `NEEDS_CONTEXT` instead of guessing when required context is missing.
 
 ---
 

package/src/gdskills/bundled/skills/review/review-style/SKILL.md

@@ -229,7 +229,7 @@ boundary is the trigger-and-outcome test.
 
 ## Orchestrated Review Contract
 
-When dispatched by `review-orchestrator`, follow the provided `reviewer-input.schema.json` payload. Return a `REVIEW_RESULT` object compatible with `skills/review-orchestrator/reviewer-finding.schema.json`, then a concise markdown summary. Keep findings evidence-based, include concrete `suggested_fix` for every blocker/major, and return `NEEDS_CONTEXT` instead of guessing when required context is missing.
+When dispatched by `review-orchestrator`, follow the provided `reviewer-input.schema.json` payload. Return a `REVIEW_RESULT` object compatible with `skills/review/review-orchestrator/reviewer-finding.schema.json`, then a concise markdown summary. Keep findings evidence-based, include concrete `suggested_fix` for every blocker/major, and return `NEEDS_CONTEXT` instead of guessing when required context is missing.
 
 ---
 

package/src/gdskills/bundled/skills/review/review-testing-practices/SKILL.md

@@ -109,7 +109,7 @@ boundary is the trigger-and-outcome test.
 
 ## Orchestrated Review Contract
 
-When dispatched by `review-orchestrator`, follow the provided `reviewer-input.schema.json` payload. Return a `REVIEW_RESULT` object compatible with `skills/review-orchestrator/reviewer-finding.schema.json`, then a concise markdown summary. Keep findings evidence-based, include concrete `suggested_fix` for every blocker/major, and return `NEEDS_CONTEXT` instead of guessing when required context is missing.
+When dispatched by `review-orchestrator`, follow the provided `reviewer-input.schema.json` payload. Return a `REVIEW_RESULT` object compatible with `skills/review/review-orchestrator/reviewer-finding.schema.json`, then a concise markdown summary. Keep findings evidence-based, include concrete `suggested_fix` for every blocker/major, and return `NEEDS_CONTEXT` instead of guessing when required context is missing.
 
 ---
 

package/src/gdskills/bundled/skills/review/review-verifier/SKILL.md

@@ -201,7 +201,7 @@ STATUS: DONE | NEEDS_CONTEXT | BLOCKED
 There is no `DONE_WITH_CONCERNS`: a verifier has no concerns of its own.
 
 Return one object conforming to
-`skills/review-orchestrator/verification-claim.schema.json`:
+`skills/review/review-orchestrator/verification-claim.schema.json`:
 
 ````text
 ```json keryx:verifications

package/src/gdskills/bundled/skills/shared/git-merge-base.md

@@ -2,7 +2,7 @@
 
 ## Purpose
 Reusable bash script for determining the review scope (merge-base) for review skills.
-Used by: code-ai-review, code-boss-review, code-mobx-store-review, code-style-review.
+Used by: code-ai-review, code-learned-review, code-mobx-store-review, code-style-review.
 
 ## Script