Tasks: plugin-validator UX and coverage gaps

Tasks: plugin-validator UX and coverage gaps

Feature Summary

Fix UX bugs and expand validation coverage in plugin-validator.py. Four independent sub-issues from groomed backlog item.

Pre-existing Test State

18 failed, 260 passed, 1 skipped as of 2026-02-14. Do NOT regress any passing tests. Do NOT fix pre-existing failures unless they are directly in the scope of a task below.

Context Manifest

Primary Files

• ./plugins/plugin-creator/scripts/plugin-validator.py (3045 lines)
• ./plugins/plugin-creator/tests/conftest.py (341 lines)

Architecture

• Validator Protocol (line 249): validate(path) -> ValidationResult, can_fix() -> bool, fix(path) -> list[str]
• FileType enum (line 138): SKILL, AGENT, COMMAND, PLUGIN, UNKNOWN
• Error codes (lines 66-112): FM001-FM010, SK001-SK007, LK001-LK002, PD001-PD003, PL001-PL005, NR001-NR002
• Validator selection (lines 2885-2925): Skills/Agents/Commands get FrontmatterValidator, NameFormatValidator, DescriptionValidator, NamespaceReferenceValidator. Skills additionally get ComplexityValidator, InternalLinkValidator, ProgressiveDisclosureValidator. Plugins get PluginStructureValidator.
• Result collection (lines 2928-2931): Iterates validators (not files), creating duplicate path entries
• Report display (lines 2620-2710): Prints per-result, counts results as files

Test Conventions

• Module loaded via importlib.util.spec_from_file_location("plugin_validator", ...) (conftest.py)
• Fixtures: cli_runner, sample_skill_dir, sample_agent_dir, sample_plugin_dir, mock_frontmatter_file
• Pattern: create tmp files with frontmatter, instantiate validator, call .validate(), assert on ValidationResult
• Class-based test organization, pytest.mark.parametrize

Linting

• Ruff py311 strict mode
• Per-file ignores for plugins/plugin-creator/scripts/*.py: B008, TC003
• Per-file ignores for **/scripts/**: ANN401, DOC, PLC0415, PLR0911, S, T201
• Per-file ignores for **/tests/**: ANN, D, DOC, E501, EXE, N, PLC, PLR, S, SLF, T

Task 1: Fix report counting — unique files not validator invocations

Status: ✅ COMPLETE Started: 2026-02-14T16:21:00Z Completed: 2026-02-14T16:30:00Z Priority: P0 (Must Have) Estimated scope: ~50 lines changed

Problem

Lines 2928-2931 iterate over validators (not files), appending (path, result) for each validator. When 1 file has 4 validators, results list has 4 entries pointing to the same path. The report at line 2630 prints "PASSED" 4 times for 1 file. The summary at line 2702 shows "Total files: 4" for 1 file.

Changes Required

1. Restructure result collection (lines 2928-2931) to group results by file path
2. Update report display (lines 2620-2710) to:
   • Show each file once with all validator results beneath it
   • Label each validator result with the validator class name (e.g., "FrontmatterValidator: PASSED")
   • A file passes overall only if ALL its validators pass
3. Update summary (line 2702) to count unique file paths

Acceptance Criteria

• Running validator on 1 file shows "Total files: 1"
• Each validator result is labeled with the validator name
• Summary counts unique files, not validator invocations
• All currently-passing tests remain passing

Test Requirements

• Add test: single file with multiple validators shows "Total files: 1"
• Add test: validator names appear in per-file output

Task 2: File-type-aware DescriptionValidator — SK004/SK005 scoping

Status: ✅ COMPLETE Started: 2026-02-14T16:31:00Z Completed: 2026-02-14T16:40:00Z Priority: P0 (Must Have) Estimated scope: ~80 lines changed

Problem

DescriptionValidator.validate() (line 1790) applies SK004 ("description too short") and SK005 ("missing trigger phrases") to all file types. Commands have a different frontmatter schema and don't need trigger phrases.

Changes Required

1. Add file_type: FileType parameter to DescriptionValidator.__init__() (or pass FileType to validate())
2. SK005 ("missing trigger phrases"): Only fire on SKILL files
3. SK004 ("description too short"): Fire on SKILL and AGENT files, not COMMAND
4. Update validator selection (lines 2885-2925) to pass FileType when constructing DescriptionValidator
5. Define new error code series CM001+ for command-specific description checks (future — stub only)

Acceptance Criteria

• SK005 only fires on SKILL files (not COMMAND or AGENT)
• SK004 fires on SKILL and AGENT files but not COMMAND
• Validator selection passes FileType context to DescriptionValidator
• All currently-passing tests remain passing

Test Requirements

• Add test: command file does NOT receive SK005 warning
• Add test: command file does NOT receive SK004 warning
• Add test: agent file does NOT receive SK005 warning
• Add test: agent file still receives SK004 warning
• Add test: skill file still receives both SK004 and SK005

Task 3: Hook validation — FileType.HOOK and HookValidator

Status: ✅ COMPLETE Started: 2026-02-14T16:41:00Z Completed: 2026-02-14T16:55:00Z Priority: P1 (Should Have) Estimated scope: ~200 lines new code

Problem

FileType enum (line 138) has no HOOK variant. detect_file_type() returns UNKNOWN for .js files in hooks/ directories and hooks.json files.

Changes Required

1. Add HOOK_SCRIPT and HOOK_CONFIG to FileType enum (line 138)
2. Update detect_file_type() to recognize:
   • .js files in hooks/ directories -> HOOK_SCRIPT
   • hooks.json files -> HOOK_CONFIG
3. Define new error code constants: HK001 (invalid hooks.json), HK002 (invalid event type), HK003 (invalid hook structure)
4. Create HookValidator class implementing Validator protocol:
   • For HOOK_CONFIG: validate JSON structure, valid event types, valid matcher patterns
   • For HOOK_SCRIPT: validate file is valid JavaScript (basic syntax check), has proper shebang
5. Update validator selection to assign HookValidator for HOOK_SCRIPT and HOOK_CONFIG files
6. Reference: /plugin-creator:claude-hooks-reference-2026 skill for hooks.json schema

Acceptance Criteria

• detect_file_type() recognizes .js files in hooks/ directories
• detect_file_type() recognizes hooks.json files
• HookValidator validates hooks.json structure
• New HK001-HK003 error codes defined and used
• All currently-passing tests remain passing

Test Requirements

• New test file: test_hook_validator.py
• Test: hooks.json with valid structure passes
• Test: hooks.json with invalid JSON fails (HK001)
• Test: hooks.json with invalid event type fails (HK002)
• Test: .js hook file detected as HOOK_SCRIPT
• Test: hooks.json detected as HOOK_CONFIG
• Test: detect_file_type for hook files returns correct FileType

Task 4: Remove dead code — nested skill error message references

Status: ✅ COMPLETE Started: 2026-02-14T16:15:00Z Completed: 2026-02-14T16:20:00Z Priority: P2 (Could Have) Estimated scope: ~10 lines changed

Problem

CORRECTION from backlog: Lines 904-911 are NOT dead code. The nested pattern plugins/{plugin}/skills/*/{name}/SKILL.md exists in the repo (e.g., python3-development/skills/testing/*/SKILL.md, python3-development/skills/development/*/SKILL.md).

However, the error message strings at lines 758-760 and 770-773 reference the nested pattern with a glob wildcard (skills/*/) which is confusing in user-facing error messages. These should be rephrased for clarity.

Changes Required

1. Update error message at lines 758-760 to clarify the nested pattern (e.g., "plugins/{plugin}/skills/{name}/SKILL.md or plugins/{plugin}/skills/{category}/{name}/SKILL.md")
2. Update error message at lines 770-773 similarly
3. Verify no test depends on exact error message text (grep tests for the old strings)

Acceptance Criteria

• Error messages clearly describe both flat and nested skill paths
• No wildcard globs in error messages shown to users
• All currently-passing tests remain passing

Test Requirements

• Verify existing NR tests still pass after message change
• No new tests needed (cosmetic change)

Dependencies

Tasks 1-4 are independent and can be implemented in parallel. Task 3 is the largest (new validator class, new error codes, new test file).

Execution Order (suggested)

1. Task 4 (smallest, cosmetic) — builds confidence, fast win
2. Task 1 (UX fix) — most visible improvement
3. Task 2 (file-type awareness) — false positive elimination
4. Task 3 (hook validation) — largest, new feature