Merged
Conversation
Create the NuGet release-notes skill with per-skill evals plus an extracted package release-notes format reference. The skill targets .NET repos that keep cumulative .nuget/<ProjectName>/PackageReleaseNotes.txt files and turns git history into per-package release notes.
Make git-visual-commits treat short prose bodies that are wrapped mid-sentence as a verification failure that must be repaired before success is reported. Add a targeted eval for repairing wrapped commit bodies while preserving bot identity and re-checking the stored message after amend.
Create a git-aware skill for writing package-facing NuGet READMEs from real project metadata, git history, and source-backed capability cues. The skill keeps branding repo-derived, uses associated tests as supporting hints when feasible, and ships with per-skill evals plus a README blueprint reference.
Add a runner-agnostic overlay for Anthropic skill-creator so repo-managed skill work keeps temp-workspace isolation, valid benchmark layout, and honest measured-versus-simulated reporting across Codex, Copilot, and Opus.
Clarify that plain git bot commit requests apply to the entire worktree unless the user explicitly narrows scope. Add eval coverage so yolo mode still groups the full diff instead of silently committing only the latest task slice.
Document the mandatory with_skill versus without_skill benchmark flow for repo-managed skills and bring the README catalog up to date with the current skill set and commit-behavior guidance.
Review Summary by QodoAdd three new skills and enforce mandatory benchmarking workflow with human review
WalkthroughsDescription• Add three new skills: git-nuget-release-notes, git-nuget-readme, and skill-creator-agnostic with full documentation and evals • Enforce mandatory paired with_skill/without_skill benchmarking workflow for all repo-managed skills with human review via benchmark.json and eval-viewer • Refactor git-visual-commits to treat full worktree as default scope and repair mid-sentence wrapped commit bodies before success • Update documentation in AGENTS.md, CONTRIBUTING.md, and README.md to reflect stricter benchmarking requirements and new skills Diagramflowchart LR
A["New Skills"] -->|git-nuget-release-notes| B["Per-package release notes"]
A -->|git-nuget-readme| C["Package-facing README"]
A -->|skill-creator-agnostic| D["Cross-runner benchmarking overlay"]
E["Benchmarking Workflow"] -->|with_skill vs without_skill| F["Paired comparison"]
F -->|aggregate_benchmark.py| G["benchmark.json"]
G -->|generate_review.py| H["Human review artifact"]
I["git-visual-commits Refactor"] -->|Full worktree default| J["Scope clarity"]
I -->|Body repair| K["Mid-sentence wrap detection"]
File Changes1. AGENTS.md
|
Code Review by Qodo
1.
|
There was a problem hiding this comment.
Pull request overview
This PR tightens the documented benchmarking/evaluation workflow for repo-managed skills (paired with_skill/without_skill runs from a temp workspace with human review via benchmark.json + eval viewer), and adds three new skills with supporting references and eval definitions.
Changes:
- Document stricter benchmark workflow requirements across
README.md,CONTRIBUTING.md, andAGENTS.md. - Add new skills:
git-nuget-release-notes,git-nuget-readme, andskill-creator-agnostic(each withSKILL.md, references, and per-skill evals). - Expand
git-visual-commitsguidance and evals around default scope and commit-body verification/repair.
Reviewed changes
Copilot reviewed 15 out of 15 changed files in this pull request and generated 2 comments.
Show a summary per file
| File | Description |
|---|---|
| skills/skill-creator-agnostic/SKILL.md | New skill overlay documenting runner-agnostic skill creation/benchmark guardrails. |
| skills/skill-creator-agnostic/references/windows-powershell-benchmarking.md | New Windows/PowerShell benchmark hygiene reference (encoding, paths, etc.). |
| skills/skill-creator-agnostic/references/benchmark-contract.md | New benchmark artifact/layout contract reference for reproducible runs. |
| skills/skill-creator-agnostic/evals/evals.json | New per-skill eval prompts/expectations for the overlay skill. |
| skills/git-visual-commits/SKILL.md | Clarifies default full-worktree scope and commit-body wrap verification/repair behavior. |
| skills/git-visual-commits/evals/evals.json | Adds evals covering mid-sentence wrap repair and full-worktree default scope in yolo mode. |
| skills/git-nuget-release-notes/SKILL.md | New skill for generating per-package PackageReleaseNotes.txt from git + project metadata. |
| skills/git-nuget-release-notes/references/package-release-notes-format.md | New reference blueprint for the normalized PackageReleaseNotes format/sections. |
| skills/git-nuget-release-notes/evals/evals.json | New per-skill eval prompts/expectations for release-notes generation behavior. |
| skills/git-nuget-readme/SKILL.md | New skill for updating NuGet-facing README.md grounded in git + metadata + source/tests. |
| skills/git-nuget-readme/references/nuget-readme-blueprint.md | New reference blueprint for a NuGet adoption-focused README structure/tone. |
| skills/git-nuget-readme/evals/evals.json | New per-skill eval prompts/expectations for README update behavior. |
| README.md | Documents new benchmark workflow requirements and lists/introduces the new skills. |
| CONTRIBUTING.md | Adds paired benchmark + human review requirements to contribution guidance/checklist. |
| AGENTS.md | Reinforces temp-workspace eval execution + paired benchmarking + artifact review guidance. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
README.md
Outdated
Comment on lines
+17
to
+21
| Validation follows the same philosophy: run | ||
| `scripts/validate-skill-templates.ps1` locally for the fast feedback | ||
| loop, and let GitHub Actions rerun that same script on pull requests as | ||
| the safety net. That validator also checks skill frontmatter metadata | ||
| such as the 1024-character YAML description limit. | ||
| such as the 1024-character YAML description limit and the presence of the repo's benchmark-workflow policy in the expected docs. |
Comment on lines
+4
to
+9
| Adds runner-agnostic guardrails on top of Anthropic's skill-creator | ||
| for creating, modifying, and benchmarking skills across Codex, GitHub | ||
| Copilot, Opus, and similar agents. Use whenever skill work must | ||
| follow temp-workspace isolation, valid `eval-N/<config>/run-N/` | ||
| benchmark layout, honest measured-vs-simulated labeling, UTF-8-safe | ||
| artifact generation, and repo-managed skill sync/README update rules. |
![qodo-code-review[bot]](https://avatars.githubusercontent.com/in/484649?s=60&v=4){kind=small}
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 6d2e6e78fb
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| Typical flow: | ||
|
|
||
| ```powershell | ||
| python "$HOME/.agents/skills/skill-creator/scripts/aggregate_benchmark.py" ` |
There was a problem hiding this comment.
These command examples hard-code "$HOME/.agents/skills/skill-creator/...", which breaks the advertised runner-/environment-agnostic workflow when skill-creator is installed elsewhere (for example only under ~/.claude/skills or a system path). In those setups the documented aggregation/review steps fail with file-not-found, so users cannot produce benchmark.json/review artifacts even if their benchmark layout is correct; use a discovered skill path (or an explicit variable) instead of a fixed install location.
Useful? React with 👍 / 👎.
Standardizes documentation across AGENTS.md, CONTRIBUTING.md, and README.md to explicitly instruct users to resolve the installed Anthropic skill-creator path (typically under ~/.agents/skills/skill-creator/ or ~/.claude/skills/skill-creator/) before running benchmark and review tools. Replaces implicit path assumptions with explicit resolver guidance to support both local install locations.
Corrects placeholder syntax from <ProjectName> to {ProjectName} for proper YAML semantics throughout the skill documentation. Renames 'Non-Negotiable Rules' section to 'Critical' for consistency with other repo-managed skills and improved clarity of critical requirements.
Updates benchmark directory pattern from eval-N to iteration-N/eval-name/{config}/run-N/ for clarity and consistency. Adds PowerShell resolver logic to benchmark-contract.md that probes ~/.agents/skills/skill-creator/ and ~/.claude/skills/skill-creator/, then runs benchmark and review scripts from the resolved root. Updates SKILL.md to clarify that users must resolve the installed skill-creator path before calling benchmark tools. Renames 'Non-Negotiable Rules' to 'Critical' for consistency across repo skills.
Rewraps long lines in AGENTS.md, CONTRIBUTING.md, CHANGELOG.md, and README.md for improved readability and consistent line length. Removes extra blank lines and normalizes list formatting across repo governance and policy documents.
Normalizes line wrapping and formatting across all SKILL.md, FORMS.md, and references/ files. Compacts multi-line YAML descriptions and removes extra blank lines for consistent documentation presentation. Improves readability while preserving all content and functional meaning.
Normalizes line wrapping in shared asset templates including .github/copilot-instructions.md, asset CHANGELOG.md bootstrap files, and nuget-readme.md package documentation templates. Removes extra blank lines and improves readability while preserving all content.
Adds the v0.3.1 release entry (2026-03-19) documenting three new skills (git-nuget-release-notes, git-nuget-readme, skill-creator-agnostic) and seven key improvements to git-visual-commits, benchmark tooling, skill formatting, and repo documentation. Updates compare links to reflect the new release version and keeps the Unreleased placeholder ready for future work.
Adds explicit support for release-intent trigger words (finalize, release, publish, ship) that automatically extract and use the version from version-prefixed branch names (e.g., v0.3.1/additional-skills becomes 0.3.1). Updates skill description to advertise the new triggers and clarifies Step 2 workflow to prioritize release-intent words when determining changelog target section. This streamlines release finalization by eliminating the need to manually specify versions.
Condenses the skill description from 1044 to 667 characters to comply with the repo's 1024-character limit while preserving all trigger phrases (finalize, ready to release, rtr, release) and key capabilities. Tightens wording without losing clarity about when to use the skill or what it does.
Completes the v0.3.1 release entry with curated narrative highlighting three new NuGet-focused skills (git-nuget-release-notes, git-nuget-readme, skill-creator-agnostic) and eight key improvements to repo automation, documentation standards, and skill refinements. Release classified as patch (docs, formatting, skill improvements; no breaking changes).
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This pull request updates documentation to require a stricter and more transparent benchmarking workflow for all repo-managed skills, and introduces three new skills with supporting documentation and usage examples. The most important changes are grouped below by theme.
Benchmarking and Evaluation Workflow Improvements:
with_skillandwithout_skillruns before work is considered complete. For new skills, the baseline iswithout_skill; for existing skills, it can be eitherwithout_skillor the previous/original version, matching theskill-creatorbenchmark flow. Human review of results viabenchmark.jsonandeval-viewer/generate_review.pyis now required, and generated artifacts must not be committed unless explicitly requested. These requirements are documented inREADME.md,CONTRIBUTING.md, andAGENTS.md. [1] [2] [3] [4]New Skills Added and Documented:
git-nuget-release-notes,git-nuget-readme, andskill-creator-agnostic. Each skill is now listed in the skills table and has a dedicated usage section in theREADME.md. [1] [2] [3]Skill Workflow and Validation Enhancements:
scripts/validate-skill-templates.ps1) supplement, but do not replace, per-skill evals, and that validation includes checking for the presence of the benchmark workflow policy. [1] [2]Improvements to Existing Skill Documentation:
git-visual-commitsdocumentation now clarifies that by default, all files in the worktree are committed and that mid-sentence wrapping in commit bodies is treated as a verification failure, with automatic repair.These changes ensure a more rigorous, reviewable, and reproducible workflow for skill development and introduce new capabilities for .NET package documentation and benchmarking.