swift-code-reviewer-skill 1.2.1 → 1.4.0

package/CHANGELOG.md

@@ -7,6 +7,57 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [1.4.0] - 2026-05-07
+
+### Added
+
+- **Multi-agent installer** — `npx swift-code-reviewer-skill init` now prompts to select one or more
+  agent targets (Claude Code, OpenAI Codex CLI, Google Gemini CLI, Kiro) via an interactive
+  `@inquirer/prompts` checkbox; `--agent <name[,name]>` and `--all` flags bypass the prompt for
+  CI/scripted use; non-TTY without flags defaults to `claude` (preserves all existing CI scripts)
+- **Canonical core** (`core/swift-code-reviewer.core.md`) — agent-agnostic source of truth with
+  `<PROJECT_STANDARDS_FILE>` and `<COMPANION_REF:...>` tokens; per-agent wrappers substitute back
+- **Per-agent sidecar templates** under `templates/agents/{claude,codex,gemini,kiro}/`:
+  - `claude/swift-code-reviewer.md` — loads from `~/.claude/skills/` with full companion-skill refs
+  - `codex/swift-code-reviewer.md` — standalone, references `AGENTS.md`, condensed rules inlined
+  - `gemini/swift-code-reviewer.md` — standalone, references `GEMINI.md`, condensed rules inlined
+  - `kiro/swift-code-reviewer.md` — `inclusion: fileMatch` + `fileMatchPattern: "**/*.swift"`,
+    auto-activates on any Swift file; workspace-scoped (global steering has known bug #6171)
+- **Per-agent command templates** under `templates/commands/{claude,gemini}/`:
+  - `claude/review.md` — `/review` slash command (unchanged behavior)
+  - `gemini/review.toml` — Gemini custom command; `prompt` uses `@./swift-code-reviewer.md`
+- **`examples/`** directory — three real review reports against pinned OSS PRs (cited by commit SHA)
+- **`assets/init-demo.tape`** — VHS tape script to regenerate the interactive-install GIF
+- `bin/lib/agents.js` — idempotent per-agent install functions (Created / Updated / Skipped logging)
+- `bin/lib/prompt.js` — TTY-aware agent selection (`selectAgents`)
+- `__tests__/installer.test.js` — installer tests using Node's built-in `node:test`
+- `CONTRIBUTING.md` "Adding a new agent target" section documents the contribution path
+
+### Changed
+
+- `bin/install.js` refactored: argv parse for flags, delegates to `bin/lib/agents.js` and
+  `bin/lib/prompt.js`; post-install hint is now per-agent (Codex shows AGENTS.md snippet,
+  Gemini shows TOML command hint, Kiro notes workspace-scoped steering)
+- `templates/agents/swift-code-reviewer.md` and `templates/commands/review.md` moved to
+  `templates/agents/claude/` and `templates/commands/claude/` respectively
+- README fully rewritten: monozukuri spine, per-agent capability matrix, per-agent quirks section,
+  real review excerpt replacing synthetic `LoginViewModel.swift` block
+
+### Notes
+
+- New runtime dependency: `@inquirer/prompts@^7` (~200 KB added to install size)
+- Codex does not support `/review` slash commands; `@`-mentions in `AGENTS.md` are not
+  auto-resolved — the post-install hint provides the exact snippet to paste
+- Kiro global-steering has a known bug ([#6171](https://github.com/kirodotdev/Kiro/issues/6171));
+  the installer uses workspace-scoped `fileMatch` steering instead
+
+## [1.3.0] - 2026-05-07
+
+### Added
+
+- **Spec adherence review** (`references/spec-adherence.md`) — validates implementation against PR description and linked issues, flagging scope drift and unimplemented requirements
+- **Agent-loop feedback** (`references/agent-loop-feedback.md`) — meta-review layer that identifies recurring patterns suggesting gaps in the agent's own instructions, improving future AI-generated code quality
+
 ## [1.2.1] - 2026-04-21
 
 ### Fixed
@@ -113,6 +164,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## Version History Summary
 
+- **1.4.0** (2026-05-07): Multi-agent installer (Claude/Codex/Gemini/Kiro), canonical core, per-agent sidecar templates, real examples, VHS tape
+- **1.3.0** (2026-05-07): Add spec adherence review and agent-loop meta-feedback layer
 - **1.2.1** (2026-04-21): Fix installer not copying `skills/` and `templates/` directories
 - **1.2.0** (2026-04-21): Bundle five companion Swift skills, add `init` scaffolding command, skill-review CI action, SKILL.md condensed 71%
 - **1.1.1** (2026-03-24): Fix incorrect `install-skill.sh` (was XcodeBuildMCP installer)

package/CONTRIBUTING.md

@@ -33,6 +33,7 @@ We welcome suggestions for:
 
 1. Fork the repository
 2. Clone your fork:
+
    ```bash
    git clone https://github.com/yourusername/swift-code-reviewer-skill.git
    cd swift-code-reviewer-skill
@@ -54,7 +55,7 @@ When modifying the main skill file:
 3. Update table of contents if adding sections
 4. Test with real Swift/SwiftUI code
 
-**Reference Files (references/*.md)**
+**Reference Files (references/\*.md)**
 
 When updating reference documentation:
 
@@ -79,23 +80,28 @@ When updating reference documentation:
 
 Always use this format for examples:
 
-```markdown
+````markdown
 **Check for:**
+
 - [ ] Pattern or rule to check
 
 **Examples:**
 
 ❌ **Bad: [Description]**
+
 ```swift
 // Code showing the anti-pattern
 ```
+````
 
 ✅ **Good: [Description]**
+
 ```swift
 // Code showing the correct pattern
 ```
 
 **Why This Matters**: [Explanation of impact]
+
 ```
 
 #### Testing Your Changes
@@ -123,11 +129,13 @@ Always use this format for examples:
 Write clear, descriptive commit messages:
 
 ```
+
 Add force unwrap detection in security checklist
 
 - Added pattern detection for !, as!, and try!
 - Included examples of safe alternatives
 - Updated security-checklist.md with new section
+
 ```
 
 **Format:**
@@ -150,6 +158,78 @@ Add force unwrap detection in security checklist
 5. **Respond to feedback** from reviewers
 6. **Keep commits clean** (squash if needed)
 
+## Adding a New Agent Target
+
+The skill ships with four built-in targets (Claude, Codex, Gemini, Kiro). Adding another
+follows a four-step checklist:
+
+### 1. Create the agent template
+
+```
+
+templates/agents/<agent>/swift-code-reviewer.md
+
+```
+
+Base it on `core/swift-code-reviewer.core.md`. Substitute:
+- `<PROJECT_STANDARDS_FILE>` → the path your agent reads for project standards
+- `<COMPANION_REF:...>` → inline the key excerpt or omit if the agent can read
+  the file at `~/.claude/skills/swift-code-reviewer-skill/…`
+
+Add agent-specific frontmatter if required (e.g., Kiro's `inclusion: fileMatch`).
+
+If your agent supports slash commands, also create:
+```
+
+templates/commands/<agent>/review.<ext>
+
+````
+
+### 2. Add an install function in `bin/lib/agents.js`
+
+```js
+function installMyAgent(packageRoot, cwd, { dryRun = false, force = false } = {}) {
+  // copy template(s), print post-install hints
+}
+````
+
+Export it and add it to `AGENT_INSTALLERS`:
+
+```js
+const AGENT_INSTALLERS = {
+  claude: installClaude,
+  codex: installCodex,
+  gemini: installGemini,
+  kiro: installKiro,
+  myagent: installMyAgent, // ← add here
+};
+```
+
+### 3. Update the README
+
+- Add a row to the **Capability matrix** table.
+- Add a subsection under **Per-agent quirks & limitations** documenting
+  any behavioral differences or known issues.
+
+### 4. Add tests in `__tests__/installer.test.js`
+
+Cover at minimum: file creation, idempotency (skip on second run), dry-run
+creates no files, and any frontmatter assertions specific to your agent.
+
+### Checklist
+
+- [ ] `templates/agents/<agent>/swift-code-reviewer.md` created
+- [ ] Command template created (if supported)
+- [ ] `bin/lib/agents.js` `installMyAgent` function added and exported
+- [ ] `AGENT_INSTALLERS` map updated
+- [ ] `VALID_AGENTS` implicitly updated (it's `Object.keys(AGENT_INSTALLERS)`)
+- [ ] README capability matrix row added
+- [ ] README per-agent quirks section added
+- [ ] Tests added in `__tests__/installer.test.js`
+- [ ] `CHANGELOG.md` updated
+
+---
+
 ## What to Contribute
 
 ### High Priority
@@ -198,7 +278,7 @@ Add force unwrap detection in security checklist
 - **Clear and concise**: Avoid unnecessary words
 - **Active voice**: "Use @Observable" not "@ Observable should be used"
 - **Specific**: "View body exceeds 50 lines" not "View is too large"
-- **Educational**: Explain *why*, not just *what*
+- **Educational**: Explain _why_, not just _what_
 - **Respectful**: Assume good intentions, be constructive
 
 ## Review Criteria
@@ -214,6 +294,7 @@ Pull requests are evaluated on:
 ## Recognition
 
 Contributors will be:
+
 - Listed in the README
 - Credited in release notes
 - Thanked in the community
@@ -232,6 +313,7 @@ Significant contributions may result in becoming a maintainer.
 ### Our Pledge
 
 We are committed to providing a welcoming and inclusive environment for all contributors, regardless of:
+
 - Experience level
 - Background
 - Identity
@@ -256,6 +338,7 @@ We are committed to providing a welcoming and inclusive environment for all cont
 ### Enforcement
 
 Violations will result in:
+
 1. Warning
 2. Temporary ban
 3. Permanent ban (severe or repeated violations)