Restore /look:again and /look:tidy under /look: namespace

.claude-plugin/marketplace.json

@@ -11,13 +11,14 @@
       "name": "look",
       "source": "./src",
       "description": "Sequential code review with fresh agent contexts. Runs multiple independent review passes to catch more issues.",
-      "version": "0.5.0",
+      "version": "0.5.1",
       "author": { "name": "HartBrook" },
       "repository": "https://github.com/HartBrook/lookagain",
       "license": "MIT",
       "keywords": ["code-review", "quality", "iterative", "multi-pass"],
+      "commands": ["./commands/again.md", "./commands/tidy.md"],
       "agents": ["./agents/lookagain-reviewer.md"],
-      "skills": ["./skills/again", "./skills/tidy", "./skills/lookagain-output-format"],
+      "skills": ["./skills/lookagain-output-format"],
       "strict": false
     }
   ]

CHANGELOG.md

@@ -5,6 +5,14 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [0.5.1] - 2026-05-12
+
+### Changed
+
+- Migrated `/look:again` and `/look:tidy` back from the `skills/` directory format to the `commands/` directory format so they appear under the `/look:` namespace in the slash command picker (matching the README) and surface their description chip when typed. The underlying multi-command misrouting bug that motivated the 0.4.0 migration was fixed upstream in Claude Code 2.1.101 (release notes: "Fixed several plugin issues: slash commands resolving to the wrong plugin with duplicate `name:` frontmatter ..."), so the workaround is no longer needed. As a defense-in-depth measure, the `name:` frontmatter field is now omitted from both command files - the slash name is derived from the filename, which removes any possibility of a duplicate-name collision regardless of the resolver version.
+- The `lookagain-output-format` skill remains under `src/skills/` because the reviewer subagent invokes it via the Skill tool; it is not a user-facing slash command and would break if moved to `commands/`.
+- Updated `scripts/test.sh`, `scripts/package.sh`, `evals/promptfooconfig.yaml`, and `CONTRIBUTING.md` to reflect the new file layout.
+
 ## [0.5.0] - 2026-05-12
 
 ### Added

CONTRIBUTING.md

@@ -123,26 +123,29 @@ You can also test the plugin through the marketplace install flow, which is clos
 
 ### Key Files
 
-- **[src/skills/again/SKILL.md](src/skills/again/SKILL.md)**: Main orchestrator logic. Controls pass execution, auto-fixing, and aggregation.
-- **[src/skills/tidy/SKILL.md](src/skills/tidy/SKILL.md)**: Tidy skill for pruning old review runs.
+- **[src/commands/again.md](src/commands/again.md)**: Main orchestrator logic. Controls pass execution, auto-fixing, and aggregation.
+- **[src/commands/tidy.md](src/commands/tidy.md)**: Tidy command for pruning old review runs.
 - **[src/agents/lookagain-reviewer.md](src/agents/lookagain-reviewer.md)**: Reviewer subagent. Defines how individual review passes work.
-- **[src/skills/lookagain-output-format/SKILL.md](src/skills/lookagain-output-format/SKILL.md)**: JSON output format specification.
+- **[src/skills/lookagain-output-format/SKILL.md](src/skills/lookagain-output-format/SKILL.md)**: JSON output format specification (invoked by the reviewer subagent via the Skill tool).
 - **[src/dot-claude-plugin/plugin.json](src/dot-claude-plugin/plugin.json)**: Plugin metadata and version.
 - **[.claude-plugin/marketplace.json](.claude-plugin/marketplace.json)**: Marketplace manifest for plugin discovery and installation.
 
-### Writing Skill Prompts
+### Writing Command Prompts
 
-When editing or adding skills in `src/skills/`:
+When editing or adding commands in `src/commands/`:
 
-- Each skill is a directory containing a `SKILL.md` file (e.g., `src/skills/again/SKILL.md`).
-- **Use `$ARGUMENTS` for the raw string.** Claude Code replaces `$ARGUMENTS` with whatever the user typed after the skill name. There is no `$ARGUMENTS.name` dot-access syntax — only `$ARGUMENTS` (whole string) and `$ARGUMENTS[N]` (positional). Do NOT use an `arguments:` array in frontmatter — it is not a supported Claude Code feature and will not be interpolated.
+- Each command is a single `.md` file (e.g., `src/commands/again.md`). The slash name is derived from the filename, so `again.md` becomes `/look:again`. Do not add a `name:` field to the frontmatter - that historically caused cross-plugin name collisions (see CHANGELOG 0.5.1).
+- **Use `$ARGUMENTS` for the raw string.** Claude Code replaces `$ARGUMENTS` with whatever the user typed after the command. There is no `$ARGUMENTS.name` dot-access syntax - only `$ARGUMENTS` (whole string) and `$ARGUMENTS[N]` (positional). Do NOT use an `arguments:` array in frontmatter - it is not a supported Claude Code feature and will not be interpolated.
 - **Add `argument-hint`** in frontmatter to document expected input format (e.g., `argument-hint: "[key=value ...]"`).
-- **Add `disable-model-invocation: true`** for user-triggered actions to prevent Claude from auto-invoking them.
 - **Include a defaults table** in the body listing each key, its default, and a description. Instruct the agent to parse `key=value` pairs from `$ARGUMENTS` and fall back to defaults for missing keys.
 - **Log the resolved configuration** so it is visible in output and reviewable in evals.
-- `make test` enforces that skills using `$ARGUMENTS` have `argument-hint` in frontmatter, a defaults table in the body, and do NOT use the unsupported `arguments:` frontmatter array.
+- `make test` enforces that commands using `$ARGUMENTS` have `argument-hint` in frontmatter, a defaults table in the body, and do NOT use the unsupported `arguments:` frontmatter array.
 - After changing prompt logic, run `make eval` to verify models still interpret the arguments correctly.
 
+### Writing Skill Prompts
+
+`src/skills/` is reserved for skills invoked by other components via the Skill tool (currently only `lookagain-output-format`, which the reviewer subagent calls to get the JSON output spec). Skills are not user-invokable slash commands; if you need a new user-facing slash command, add it under `src/commands/` instead.
+
 ## Pull Requests
 
 1. Fork the repository

evals/promptfooconfig.yaml

@@ -14,7 +14,7 @@ tests:
   # ==================================================================
   - description: "no arguments → model uses all defaults (auto-fix=true, passes=3, thorough)"
     vars:
-      prompt_file: src/skills/again/SKILL.md
+      prompt_file: src/commands/again.md
     assert:
       - type: llm-rubric
         value: >
@@ -30,7 +30,7 @@ tests:
   # ==================================================================
   - description: "auto-fix=true → model plans to apply fixes"
     vars:
-      prompt_file: src/skills/again/SKILL.md
+      prompt_file: src/commands/again.md
       arg_passes: "3"
       arg_target: staged
       arg_auto-fix: "true"
@@ -45,7 +45,7 @@ tests:
 
   - description: "auto-fix=false → model skips fixes"
     vars:
-      prompt_file: src/skills/again/SKILL.md
+      prompt_file: src/commands/again.md
       arg_passes: "3"
       arg_target: staged
       arg_auto-fix: "false"
@@ -63,7 +63,7 @@ tests:
   # ==================================================================
   - description: "passes=5 → model plans exactly 5 initial passes"
     vars:
-      prompt_file: src/skills/again/SKILL.md
+      prompt_file: src/commands/again.md
       arg_passes: "5"
       arg_target: staged
       arg_auto-fix: "true"
@@ -83,7 +83,7 @@ tests:
   # ==================================================================
   - description: "model=fast → reviewer uses haiku"
     vars:
-      prompt_file: src/skills/again/SKILL.md
+      prompt_file: src/commands/again.md
       arg_passes: "3"
       arg_target: staged
       arg_auto-fix: "true"
@@ -99,7 +99,7 @@ tests:
 
   - description: "model=thorough → no explicit model override"
     vars:
-      prompt_file: src/skills/again/SKILL.md
+      prompt_file: src/commands/again.md
       arg_passes: "3"
       arg_target: staged
       arg_auto-fix: "true"
@@ -117,7 +117,7 @@ tests:
   # ==================================================================
   - description: "target=branch → branch-based diff scope"
     vars:
-      prompt_file: src/skills/again/SKILL.md
+      prompt_file: src/commands/again.md
       arg_passes: "3"
       arg_target: branch
       arg_auto-fix: "true"
@@ -135,7 +135,7 @@ tests:
   # ==================================================================
   - description: "only auto-fix=false → defaults for everything else, no fixing"
     vars:
-      prompt_file: src/skills/again/SKILL.md
+      prompt_file: src/commands/again.md
       arg_auto-fix: "false"
     assert:
       - type: llm-rubric
@@ -150,7 +150,7 @@ tests:
   # ==================================================================
   - description: "summary tables include Suggested Fix column and global ID"
     vars:
-      prompt_file: src/skills/again/SKILL.md
+      prompt_file: src/commands/again.md
     assert:
       - type: icontains
         value: "Suggested Fix"
@@ -166,7 +166,7 @@ tests:
 
   - description: "aggregate.md preserves full Suggested Fix text; conversation may truncate"
     vars:
-      prompt_file: src/skills/again/SKILL.md
+      prompt_file: src/commands/again.md
     assert:
       - type: llm-rubric
         value: >
@@ -178,7 +178,7 @@ tests:
 
   - description: "must_fix table includes a Fixed column; should_fix and suggestions do not"
     vars:
-      prompt_file: src/skills/again/SKILL.md
+      prompt_file: src/commands/again.md
     assert:
       - type: llm-rubric
         value: >
@@ -192,7 +192,7 @@ tests:
   # ==================================================================
   - description: "all=true → removes all runs"
     vars:
-      prompt_file: src/skills/tidy/SKILL.md
+      prompt_file: src/commands/tidy.md
       arg_keep: "1"
       arg_all: "true"
     assert:
@@ -203,7 +203,7 @@ tests:
 
   - description: "all=false, keep=3 → date-based retention"
     vars:
-      prompt_file: src/skills/tidy/SKILL.md
+      prompt_file: src/commands/tidy.md
       arg_keep: "3"
       arg_all: "false"
     assert:
@@ -218,7 +218,7 @@ tests:
   # ==================================================================
   - description: "tidy with no arguments → keep=1, all=false"
     vars:
-      prompt_file: src/skills/tidy/SKILL.md
+      prompt_file: src/commands/tidy.md
     assert:
       - type: llm-rubric
         value: >

scripts/package.sh

@@ -32,6 +32,7 @@ cp -r "$PROJECT_ROOT/src/dot-claude" "$DIST_DIR/lookagain/.claude"
 cp -r "$PROJECT_ROOT/src/dot-claude-plugin" "$DIST_DIR/lookagain/.claude-plugin"
 # Overlay marketplace.json from repo root into the dist .claude-plugin/ directory
 cp "$PROJECT_ROOT/.claude-plugin/marketplace.json" "$DIST_DIR/lookagain/.claude-plugin/"
+cp -r "$PROJECT_ROOT/src/commands" "$DIST_DIR/lookagain/commands"
 cp -r "$PROJECT_ROOT/src/agents" "$DIST_DIR/lookagain/agents"
 cp -r "$PROJECT_ROOT/src/skills" "$DIST_DIR/lookagain/skills"
 

scripts/test.sh

@@ -82,7 +82,7 @@ test_plugin_json() {
     fi
 
     # Required fields
-    for field in name version description author agents skills; do
+    for field in name version description author commands agents skills; do
         check_json_field "$pjson" "$field"
     done
 
@@ -111,8 +111,8 @@ sys.exit(0 if isinstance(d.get('author'), dict) and 'name' in d['author'] else 1
 
 test_required_files() {
     check_file "src/dot-claude-plugin/plugin.json"
-    check_file "src/skills/again/SKILL.md"
-    check_file "src/skills/tidy/SKILL.md"
+    check_file "src/commands/again.md"
+    check_file "src/commands/tidy.md"
     check_file "src/agents/lookagain-reviewer.md"
     check_file "src/skills/lookagain-output-format/SKILL.md"
     check_file "src/dot-claude/settings.local.json"
@@ -123,21 +123,22 @@ test_required_files() {
 }
 
 test_frontmatter() {
-    check_frontmatter "$PROJECT_ROOT/src/skills/again/SKILL.md" name description
-    check_frontmatter "$PROJECT_ROOT/src/skills/tidy/SKILL.md" name description
+    # Commands derive their name from the filename, so only description is required.
+    check_frontmatter "$PROJECT_ROOT/src/commands/again.md" description
+    check_frontmatter "$PROJECT_ROOT/src/commands/tidy.md" description
     check_frontmatter "$PROJECT_ROOT/src/agents/lookagain-reviewer.md" name description tools
     check_frontmatter "$PROJECT_ROOT/src/skills/lookagain-output-format/SKILL.md" name description
 }
 
 test_argument_handling() {
-    # Verify that skill files using arguments follow the correct pattern:
+    # Verify that command and skill files using arguments follow the correct pattern:
     # 1. Frontmatter has argument-hint (not the unsupported arguments: array)
     # 2. Body contains $ARGUMENTS placeholder for the raw argument string
     # 3. Body contains a defaults table with Key/Default columns
     # This ensures the agent receives and parses arguments at runtime
     # rather than relying on non-existent compile-time interpolation.
 
-    for file in "$PROJECT_ROOT"/src/skills/*/SKILL.md; do
+    for file in "$PROJECT_ROOT"/src/commands/*.md "$PROJECT_ROOT"/src/skills/*/SKILL.md; do
         local relpath="${file#"$PROJECT_ROOT"/}"
 
         local frontmatter
@@ -178,13 +179,13 @@ test_argument_handling() {
 }
 
 test_again_numbering_format() {
-    # Verify the again skill documents the global-ID + Suggested Fix format
+    # Verify the again command documents the global-ID + Suggested Fix format
     # added on feat/numbered-suggested-fixes. These checks catch regressions
     # if a future edit removes the numbering scheme or the Suggested Fix
     # column from the summary template.
 
-    local skill="$PROJECT_ROOT/src/skills/again/SKILL.md"
-    local relpath="src/skills/again/SKILL.md"
+    local skill="$PROJECT_ROOT/src/commands/again.md"
+    local relpath="src/commands/again.md"
 
     if [[ ! -f "$skill" ]]; then
         fail "$relpath not found"
@@ -252,6 +253,16 @@ test_again_numbering_format() {
 test_cross_references() {
     local pjson="$PROJECT_ROOT/src/dot-claude-plugin/plugin.json"
 
+    # Commands resolve as files
+    while IFS= read -r cmd; do
+        local resolved="$PROJECT_ROOT/src/${cmd#./}"
+        if [[ -f "$resolved" ]]; then
+            pass "command $cmd resolves"
+        else
+            fail "command $cmd not found at src/${cmd#./}"
+        fi
+    done < <(python3 -c "import json,sys; [print(c) for c in json.load(open(sys.argv[1])).get('commands', [])]" "$pjson")
+
     # Agents resolve
     while IFS= read -r agent; do
         local resolved="$PROJECT_ROOT/src/${agent#./}"
@@ -305,7 +316,7 @@ test_build() {
         fail "dist marketplace.json is invalid"
     fi
 
-    for f in agents/lookagain-reviewer.md skills/again/SKILL.md skills/tidy/SKILL.md skills/lookagain-output-format/SKILL.md README.md; do
+    for f in agents/lookagain-reviewer.md commands/again.md commands/tidy.md skills/lookagain-output-format/SKILL.md README.md; do
         if [[ -f "$dist/$f" ]]; then
             pass "dist/$f exists"
         else
@@ -401,7 +412,7 @@ sys.exit(0 if pv == mv else 1)
     fi
 
     # Commands, agents, skills match plugin.json
-    for field in agents skills; do
+    for field in commands agents skills; do
         if python3 -c "
 import json, sys
 p = sorted(json.load(open(sys.argv[1])).get(sys.argv[3], []))

src/skills/again/SKILL.md → src/commands/again.md

@@ -1,7 +1,5 @@
 ---
-name: again
 description: Run sequential code review passes with fresh contexts to catch more issues
-disable-model-invocation: true
 argument-hint: "[key=value ...]"
 ---
 

src/skills/tidy/SKILL.md → src/commands/tidy.md

@@ -1,7 +1,5 @@
 ---
-name: tidy
 description: Remove old lookagain review runs, keeping today's results by default
-disable-model-invocation: true
 allowed-tools: Glob, Bash(rm -rf .lookagain/????-??-??T??-??-??)
 argument-hint: "[key=value ...]"
 ---

src/dot-claude-plugin/plugin.json

@@ -1,10 +1,11 @@
 {
   "name": "look",
-  "version": "0.5.0",
+  "version": "0.5.1",
   "description": "Sequential code review with fresh agent contexts. Runs multiple independent review passes to catch more issues.",
   "author": { "name": "HartBrook" },
   "repository": "https://github.com/HartBrook/lookagain",
   "keywords": ["code-review", "quality", "iterative", "multi-pass"],
+  "commands": ["./commands/again.md", "./commands/tidy.md"],
   "agents": ["./agents/lookagain-reviewer.md"],
-  "skills": ["./skills/again", "./skills/tidy", "./skills/lookagain-output-format"]
+  "skills": ["./skills/lookagain-output-format"]
 }