@jakkrichm/create-nexus-devflow 2.0.0 → 2.0.2

package/template/.agents/skills/pr-review/SKILL.md

@@ -1,143 +1,143 @@
 ---name: pr-review
-
-description: PR Review Addon - Review a pull request using GitHub review prompt patterns and the credited 9arm-skills/scrutinize discipline.
----
-
-# Phase 55: PR Review Addon
-
-## Target: $ARGUMENTS
-
-Use this workflow when you need a structured PR review without creating or changing the PR. It can review a local diff, a PR URL or number, or a task-linked branch.
-
-In DevFlow 2.0, this remains a supporting review workflow. It should feed corrective work back into `/40-Implement` or clear the path toward `/60-Report` and then `/70-Release`.
-
-Primary behavior now lives in the `pr-review-analysis` skill. Keep this workflow as the compatibility wrapper and user-facing review prompt surface.
-
-## Prompt Sources
-
-Adapted from:
-
-- `github/QA_REVIEW_SYSTEM_PROMPT.md`
-- `github/pr_reviewer.md`
-- `github/pr_parallel_orchestrator.md`
-- `github/pr_orchestrator.md`
-- `github/pr_quality_agent.md`
-- `github/pr_logic_agent.md`
-- `github/pr_security_agent.md`
-- `github/pr_structural.md`
-- `github/pr_codebase_fit_agent.md`
-- `github/pr_finding_validator.md`
-- `github/partials/full_context_analysis.md`
-
-Additional credited discipline:
-
-- `.agents/skills/9arm-skills/scrutinize/SKILL.md`
-- Source pack: `9arm-skills`
-- Credit: `thananon/9arm-skills`
-- Upstream: https://github.com/thananon/9arm-skills
-- Adapted for: Antigravity IDE / Nexus-DevFlow
-
-## Process
-
-### 1. Load Review Context
-
-Read the available context:
-
-- PR diff or local `git diff`
-- changed filenames
-- linked stage artifacts under `devflow/runs/{ID}-*/`
-- `20-spec.md`, `30-plan.md`, `40-implement.md`, `50-verify.md`, `60-report.md`, and `70-release.md` when relevant
-- project patterns and existing code around the changed files
-- legacy JSON only when migration context is still genuinely relevant
-
-### 2. Select Review Lanes
-
-Start with the `scrutinize` lens before detailed lane review:
-
-- Intent check: should this change exist in this form?
-- Smaller alternative: can the same outcome be reached with less code, lower risk, or existing patterns?
-- Actual path trace: follow runtime behavior beyond the diff.
-- Precision check: were assumptions explicit, edits surgical, success criteria verified, and uncertainty surfaced instead of guessed?
-
-Then select only the review lanes that match the scope. Do not run every lane by default; the review should be proportional to the change.
-
-| Lane | Use When |
-| :--- | :--- |
-| Correctness / bugs | Always, unless the user explicitly asks for a narrow non-code review. |
-| Type safety | Typed code, public types, schema changes, casts, `any`, or boundary parsing changed. |
-| Maintainability | The change adds duplication, cross-file patterns, API shape, migration bridges, or technical debt. |
-| Simplicity | The change adds abstractions, indirection, configuration, generalized utilities, or complex control flow. |
-| Testability | Important logic is mixed with IO, hidden state, time, random values, constructors, globals, or hard-coded dependencies. |
-| Coverage | New or changed behavior should have tests, especially bug fixes and business logic. |
-| Docs accuracy | Commands, setup, workflows, schemas, public APIs, examples, or user-visible behavior changed. |
-| AGENTS.md adherence | An `AGENTS.md` or equivalent project instruction file exists for the changed scope. |
-| Security | Auth, authorization, secrets, user input, external data, dependencies, or persistence boundaries changed. |
-| Performance | Queries, loops, rendering, data volume, hot paths, or synchronous work changed. |
-
-When a lane is skipped, briefly note why if the omission could surprise the reader. For tiny reviews, it is acceptable to state only the lanes used.
-
-### 3. Validate Findings
-
-Before reporting a finding:
-
-- verify it against the changed code
-- check whether it affects new code or pre-existing code
-- ensure it is actionable
-- include file and line references when possible
-- avoid style-only comments unless they hide real risk
-- keep each finding in its strongest matching lane; do not report the same issue twice under different labels
-- drop findings that are only suspicions, preferences, or theoretical improvements without a concrete failure mode
-- for AGENTS.md findings, quote the exact project instruction being violated
-- for docs findings, show the mismatch between what the docs say and what the code now does
-
-### 4. Output Findings First
-
-Use code-review style:
-
-1. Findings ordered by severity
-2. Open questions or assumptions
-3. Short summary
-4. Test gaps or residual risk
-
-When saving a substantial report, include a short `Source Discipline` section crediting `9arm-skills/scrutinize` and, where useful, brief `Intent Check`, `Smaller Alternative Considered`, and `Actual Path Trace` sections. Preserve the required PR review template headings.
-
-## Output
-
-Return a PR review. Do not modify files unless the user explicitly asks for fixes.
-
-**MANDATORY RULE:** Before saving a PR review report, inspect `.agent/resources/schemas/pr_review.template.md` and use its required headings and table structure.
-
-Before reporting completion, run:
-
-Review `{report_path}` against `pr_review.template.md`, keep the required headings, and remove placeholder text before completion.
-
-Replace any placeholder or template text with concrete findings, file references, risk, and verification gaps.
-
-**MANDATORY RULE:** If a running ID is linked to this PR, always save this review report to `pr_review.md` inside that workspace directory when possible:
-
-```text
-devflow/runs/{ID}-*/pr_review.md
-```
-
-## Relationship To DevFlow 2.0
-
-- Classification: Companion command
-- Mainline status: Verification and release support command, not a numbered stage
-- Typical entry points: `/50-Verify`, `PR`, `Agent code-reviewer`
-- Typical handoff targets: `PR-Followup`, `/60-Report`, `/70-Release`, `Wiki`
-
-## Sources
-
-- `AGENTS.md`
-- `.agent/resources/schemas/pr_review.template.md`
-- Related commands: `/50-Verify`, `PR`, `PR-Followup`, `Agent`, `/60-Report`, `/70-Release`, `Wiki`
-
-## Next Workflow Recommendation
-
-- **Primary**: `/40-Implement {ID}` when review finds required fixes, or `/60-Report {ID}` when the review is clean and the run needs the final summary refreshed before release execution.
-- **Why**: PR review either creates corrective implementation work or clears the path toward final summary alignment and release-facing packaging.
-- **Alternatives**:
-  - `/70-Release` when the report is already aligned and only release execution remains
-  - `Wiki` when the review establishes a durable convention or risk pattern
-  - `PR-Followup {target}` when the PR already has comments that need classification and response
-
+
+description: PR Review Addon - Review a pull request using GitHub review prompt patterns and the credited 9arm-skills/scrutinize discipline.
+---
+
+# Phase 55: PR Review Addon
+
+## Target: $ARGUMENTS
+
+Use this workflow when you need a structured PR review without creating or changing the PR. It can review a local diff, a PR URL or number, or a task-linked branch.
+
+In DevFlow 2.0, this remains a supporting review workflow. It should feed corrective work back into `/40-Implement` or clear the path toward `/60-Report` and then `/70-Release`.
+
+Primary behavior now lives in the `pr-review-analysis` skill. Keep this workflow as the compatibility wrapper and user-facing review prompt surface.
+
+## Prompt Sources
+
+Adapted from:
+
+- `github/QA_REVIEW_SYSTEM_PROMPT.md`
+- `github/pr_reviewer.md`
+- `github/pr_parallel_orchestrator.md`
+- `github/pr_orchestrator.md`
+- `github/pr_quality_agent.md`
+- `github/pr_logic_agent.md`
+- `github/pr_security_agent.md`
+- `github/pr_structural.md`
+- `github/pr_codebase_fit_agent.md`
+- `github/pr_finding_validator.md`
+- `github/partials/full_context_analysis.md`
+
+Additional credited discipline:
+
+- `.agents/skills/9arm-skills/scrutinize/SKILL.md`
+- Source pack: `9arm-skills`
+- Credit: `thananon/9arm-skills`
+- Upstream: https://github.com/thananon/9arm-skills
+- Adapted for: Antigravity IDE / Nexus-DevFlow
+
+## Process
+
+### 1. Load Review Context
+
+Read the available context:
+
+- PR diff or local `git diff`
+- changed filenames
+- linked stage artifacts under `devflow/runs/{ID}-*/`
+- `20-spec.md`, `30-plan.md`, `40-implement.md`, `50-verify.md`, `60-report.md`, and `70-release.md` when relevant
+- project patterns and existing code around the changed files
+- legacy JSON only when migration context is still genuinely relevant
+
+### 2. Select Review Lanes
+
+Start with the `scrutinize` lens before detailed lane review:
+
+- Intent check: should this change exist in this form?
+- Smaller alternative: can the same outcome be reached with less code, lower risk, or existing patterns?
+- Actual path trace: follow runtime behavior beyond the diff.
+- Precision check: were assumptions explicit, edits surgical, success criteria verified, and uncertainty surfaced instead of guessed?
+
+Then select only the review lanes that match the scope. Do not run every lane by default; the review should be proportional to the change.
+
+| Lane | Use When |
+| :--- | :--- |
+| Correctness / bugs | Always, unless the user explicitly asks for a narrow non-code review. |
+| Type safety | Typed code, public types, schema changes, casts, `any`, or boundary parsing changed. |
+| Maintainability | The change adds duplication, cross-file patterns, API shape, migration bridges, or technical debt. |
+| Simplicity | The change adds abstractions, indirection, configuration, generalized utilities, or complex control flow. |
+| Testability | Important logic is mixed with IO, hidden state, time, random values, constructors, globals, or hard-coded dependencies. |
+| Coverage | New or changed behavior should have tests, especially bug fixes and business logic. |
+| Docs accuracy | Commands, setup, workflows, schemas, public APIs, examples, or user-visible behavior changed. |
+| AGENTS.md adherence | An `AGENTS.md` or equivalent project instruction file exists for the changed scope. |
+| Security | Auth, authorization, secrets, user input, external data, dependencies, or persistence boundaries changed. |
+| Performance | Queries, loops, rendering, data volume, hot paths, or synchronous work changed. |
+
+When a lane is skipped, briefly note why if the omission could surprise the reader. For tiny reviews, it is acceptable to state only the lanes used.
+
+### 3. Validate Findings
+
+Before reporting a finding:
+
+- verify it against the changed code
+- check whether it affects new code or pre-existing code
+- ensure it is actionable
+- include file and line references when possible
+- avoid style-only comments unless they hide real risk
+- keep each finding in its strongest matching lane; do not report the same issue twice under different labels
+- drop findings that are only suspicions, preferences, or theoretical improvements without a concrete failure mode
+- for AGENTS.md findings, quote the exact project instruction being violated
+- for docs findings, show the mismatch between what the docs say and what the code now does
+
+### 4. Output Findings First
+
+Use code-review style:
+
+1. Findings ordered by severity
+2. Open questions or assumptions
+3. Short summary
+4. Test gaps or residual risk
+
+When saving a substantial report, include a short `Source Discipline` section crediting `9arm-skills/scrutinize` and, where useful, brief `Intent Check`, `Smaller Alternative Considered`, and `Actual Path Trace` sections. Preserve the required PR review template headings.
+
+## Output
+
+Return a PR review. Do not modify files unless the user explicitly asks for fixes.
+
+**MANDATORY RULE:** Before saving a PR review report, inspect `.agent/resources/schemas/pr_review.template.md` and use its required headings and table structure.
+
+Before reporting completion, run:
+
+Review `{report_path}` against `pr_review.template.md`, keep the required headings, and remove placeholder text before completion.
+
+Replace any placeholder or template text with concrete findings, file references, risk, and verification gaps.
+
+**MANDATORY RULE:** If a running ID is linked to this PR, always save this review report to `pr_review.md` inside that workspace directory when possible:
+
+```text
+devflow/runs/{ID}-*/pr_review.md
+```
+
+## Relationship To DevFlow 2.0
+
+- Classification: Companion command
+- Mainline status: Verification and release support command, not a numbered stage
+- Typical entry points: `/50-Verify`, `PR`, `Agent code-reviewer`
+- Typical handoff targets: `PR-Followup`, `/60-Report`, `/70-Release`, `Wiki`
+
+## Sources
+
+- `AGENTS.md`
+- `.agent/resources/schemas/pr_review.template.md`
+- Related commands: `/50-Verify`, `PR`, `PR-Followup`, `Agent`, `/60-Report`, `/70-Release`, `Wiki`
+
+## Next Workflow Recommendation
+
+- **Primary**: `/40-Implement {ID}` when review finds required fixes, or `/60-Report {ID}` when the review is clean and the run needs the final summary refreshed before release execution.
+- **Why**: PR review either creates corrective implementation work or clears the path toward final summary alignment and release-facing packaging.
+- **Alternatives**:
+  - `/70-Release` when the report is already aligned and only release execution remains
+  - `Wiki` when the review establishes a durable convention or risk pattern
+  - `PR-Followup {target}` when the PR already has comments that need classification and response
+

package/template/.agents/skills/pr-review-analysis/SKILL.md

@@ -1,89 +1,89 @@
----
-name: pr-review-analysis
-description: Review pull requests or local diffs with structured finding discipline, selective review lanes, and Nexus-DevFlow report templates. Use when the team needs a substantial review without turning review into a numbered workflow stage.
----
-
-# PR Review Analysis
-
-## Overview
-
-This skill is the shared behavior layer behind `PR-Review`.
-
-It preserves the stronger PR review UX from the legacy workflow while fitting DevFlow 2.0 as a non-mainline support capability.
-
-## Related Foundation Skills
-
-This skill should reuse and align with:
-
-- `.agents/skills/9arm-skills/scrutinize/SKILL.md`
-- `.agent/resources/schemas/pr_review.template.md`
-
-## When to Use
-
-- after `/50-Verify` when the change needs structured review before `/60-Report` and `/70-Release`
-- when a local diff, branch, or pull request needs findings-first analysis
-- when the user wants review without directly editing code
-
-Do not use this skill as a substitute for implementation, debugging, or formal release packaging.
-
-## Process
-
-### 1. Load Review Context
-
-Read the minimum relevant context:
-
-- PR diff or local `git diff`
-- changed files
-- linked stage artifacts under the running workspace when available
-- nearby project patterns and affected runtime paths
-
-### 2. Apply Scrutiny First
-
-Start with the `scrutinize` discipline:
-
-- intent check
-- smaller or safer alternative
-- actual runtime path trace
-- precision and evidence check
-
-### 3. Select Review Lanes
-
-Use only the lanes that match scope:
-
-- correctness
-- maintainability
-- simplicity
-- testability and coverage
-- docs accuracy
-- AGENTS or instruction adherence
-- security
-- performance
-
-### 4. Validate Findings
-
-Before reporting a finding:
-
-- verify it against the changed code
-- confirm whether it is new or pre-existing
-- keep it actionable
-- include file and line references whenever possible
-
-### 5. Save Reusable Report
-
-When saving a substantial report, preserve:
-
-```text
-devflow/runs/{ID}-*/pr_review.md
-```
-
-Use `.agent/resources/schemas/pr_review.template.md` and replace all placeholder text with real findings, questions, risks, and evidence gaps.
-
-## Output
-
-Return:
-
-- findings ordered by severity
-- open questions or assumptions
-- residual risks or test gaps
-- where the review report was saved
-- the recommended next route such as `/40-Implement`, `/70-Release`, or `PR-Followup`
+---
+name: pr-review-analysis
+description: Review pull requests or local diffs with structured finding discipline, selective review lanes, and Nexus-DevFlow report templates. Use when the team needs a substantial review without turning review into a numbered workflow stage.
+---
+
+# PR Review Analysis
+
+## Overview
+
+This skill is the shared behavior layer behind `PR-Review`.
+
+It preserves the stronger PR review UX from the legacy workflow while fitting DevFlow 2.0 as a non-mainline support capability.
+
+## Related Foundation Skills
+
+This skill should reuse and align with:
+
+- `.agents/skills/9arm-skills/scrutinize/SKILL.md`
+- `.agent/resources/schemas/pr_review.template.md`
+
+## When to Use
+
+- after `/50-Verify` when the change needs structured review before `/60-Report` and `/70-Release`
+- when a local diff, branch, or pull request needs findings-first analysis
+- when the user wants review without directly editing code
+
+Do not use this skill as a substitute for implementation, debugging, or formal release packaging.
+
+## Process
+
+### 1. Load Review Context
+
+Read the minimum relevant context:
+
+- PR diff or local `git diff`
+- changed files
+- linked stage artifacts under the running workspace when available
+- nearby project patterns and affected runtime paths
+
+### 2. Apply Scrutiny First
+
+Start with the `scrutinize` discipline:
+
+- intent check
+- smaller or safer alternative
+- actual runtime path trace
+- precision and evidence check
+
+### 3. Select Review Lanes
+
+Use only the lanes that match scope:
+
+- correctness
+- maintainability
+- simplicity
+- testability and coverage
+- docs accuracy
+- AGENTS or instruction adherence
+- security
+- performance
+
+### 4. Validate Findings
+
+Before reporting a finding:
+
+- verify it against the changed code
+- confirm whether it is new or pre-existing
+- keep it actionable
+- include file and line references whenever possible
+
+### 5. Save Reusable Report
+
+When saving a substantial report, preserve:
+
+```text
+devflow/runs/{ID}-*/pr_review.md
+```
+
+Use `.agent/resources/schemas/pr_review.template.md` and replace all placeholder text with real findings, questions, risks, and evidence gaps.
+
+## Output
+
+Return:
+
+- findings ordered by severity
+- open questions or assumptions
+- residual risks or test gaps
+- where the review report was saved
+- the recommended next route such as `/40-Implement`, `/70-Release`, or `PR-Followup`