Comparing changes

## Changes Simplified slash_commands metadata format from detailed TOML tables to simple array: ### Before (Verbose) ```toml [[tool.metaspec.slash_commands]] name = "show-spec" description = "Display specification" source = "generic" [[tool.metaspec.slash_commands]] name = "get-template" description = "Get template" source = "generic" ``` ### After (Concise) ```toml sd_type = "generic" slash_commands = ["show-spec", "get-template"] ``` ## Rationale 1. **Consistency**: Matches cli_commands format 2. **Simplicity**: 50+ lines → ~10 lines 3. **Maintainability**: Easier to scan and update 4. **Sufficient**: Detailed info available in command files ## Updated Files **Templates**: - pyproject.toml.j2: Generate simplified format * Support custom sd_type detection (e.g., "sdm") * Output slash_commands as array **Documentation**: - metaspec-metadata-examples.md: All 5 examples updated * Example 4: Marketing → SDM system (custom SD-X) * All use array format - implement.md.j2: 4 scenario examples updated * Added Scenario 3: SDM example ## Benefits - ✅ Easier to read and maintain - ✅ Consistent with CLI commands format - ✅ 80% less boilerplate - ✅ Supports custom SD-X systems (like SDM)

… 'custom' ## Motivation Enforce consistent SD-X naming pattern: `sd<domain>` where domain is 1-3 letter abbreviation **Problem**: `"generic"` doesn't follow SD-X pattern (missing 'sd' prefix) ## Changes ### Terminology Update - ❌ Before: `sd_type = "generic"` (inconsistent) - ✅ After: `sd_type = "custom"` (clearer meaning) ### SD-X Naming Convention **Standard SD-X Systems** (follow pattern): - `sds` - **S**pec-**D**riven **S**pecification - `sdd` - **S**pec-**D**riven **D**evelopment - `sdm` - **S**pec-**D**riven **M**arketing - `sdt` - **S**pec-**D**riven **T**esting - `sdo` - **S**pec-**D**riven **O**perations **Non-SD-X**: - `custom` - Custom commands (not following SD-X workflow pattern) - `none` - No slash commands ### Updated Files **Templates**: - `pyproject.toml.j2`: Change detection from "generic" → "custom" - `implement.md.j2`: Update guidance and examples - `README.md`: Update source mapping documentation - `library/*/README.md`: Update source references **Documentation**: - `metaspec-metadata-examples.md`: * Update field descriptions * Update all 5 examples * Update detection logic table * Update use case code samples ## Benefits 1. ✅ **Consistent**: All SD-X systems follow `sd<domain>` pattern 2. ✅ **Extensible**: Clear path for new SD-X systems (sdt, sdo, etc.) 3. ✅ **Semantic**: "custom" better describes non-SD-X commands 4. ✅ **Discoverable**: Easy to identify SD-X vs custom systems ## Migration For existing speckits using `"generic"`: - Change `source: "generic"` → `source: "custom"` - Change `sd_type = "generic"` → `sd_type = "custom"` - No functional changes, pure terminology update

## Motivation Generated speckit commands can exceed 500-1000 lines (e.g., marketing-spec-kit has 4 commands >800 lines), consuming excessive tokens when AI reads them. ## Solution Add guidance in `implement.md.j2` for AI to generate navigation tables for long commands (>500 lines). ## Changes ### Added Section: "3. Optimize Long Slash Commands" **Location**: After pyproject.toml guidance (lines 433-508, 75 lines) **Content**: 1. **When**: Commands exceed 500 lines 2. **Why**: Save 70-90% token consumption 3. **How**: Add navigation guide at file start 4. **Template**: Complete navigation template with: - Section breakdown (4-6 sections, 100-300 lines each) - Line numbers and read_file usage - Token savings calculation - Typical usage patterns 5. **Guidelines**: - Threshold: 500 lines - Token estimate: ~3.5 tokens/line - Section design principles 6. **Example**: 847-line command breakdown ### Updated Navigation Table **File size**: 1084 → 1158 lines (+74 lines) **Main table**: - Section 7: 275-464 (189) → 275-525 (251 lines) +62 - Section 8-10: 465-556 → 526-617 +61 - Section 11-13: 557-681 → 618-742 +61 - Section 14-15: 682-808 → 743-869 +61 - Section 16: 809-1096 → 870-1158 +61 **Section 7 subsections**: - Python: 306-446 (141) → 306-508 (203 lines) * Now includes: command navigation guidance (74 lines) **Phase-specific**: - Added: Command Navigation (433-508, 75 lines) - Updated all phase line numbers **Token savings**: - Full file: ~3800 → ~4050 tokens - Command navigation: 75 lines (~260 tokens) → **94% savings** ## Benefits 1. ✅ **Token optimization**: 70-90% savings for long commands 2. ✅ **Consistency**: Matches MetaSpec's own command style 3. ✅ **Best practice**: Guides AI to follow precision navigation pattern 4. ✅ **Scalability**: Works for any domain (SDM, SDT, SDO, etc.) ## Example Use Cases - marketing-spec-kit: 4 commands >800 lines need navigation - Future speckits with complex workflows - Any SD-X system with detailed command guidance

## Version Updates - src/metaspec/__init__.py: 0.9.6 → 0.9.7 - pyproject.toml: 0.9.6 → 0.9.7 - CHANGELOG.md: Move Unreleased to [0.9.7] - 2025-11-22 ## Release Summary **v0.9.7 - Metadata & Navigation Enhancements** Three major improvements: 1. **Simplified slash_commands Format** (refactor) - Array format instead of TOML tables - 80% less boilerplate - Consistent with cli_commands style 2. **Standardized SD-X Naming** (refactor) - "generic" → "custom" for non-SD-X - Enforced sd<domain> pattern - Clear SD-X vs custom distinction 3. **Long Command Navigation** (feat) - Guidance for commands >500 lines - 70-90% token savings - Template + examples + guidelines Benefits: - ✅ Consistent metadata format - ✅ Clear system classification - ✅ Massive token savings - ✅ Best practices automated Commits: - 7be1944: Simplify slash_commands format - 8b2baea: Standardize SD-X naming - 38928fe: Add navigation guidance

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Comparing changes

Open a pull request

Commits on Nov 22, 2025

This comparison is taking too long to generate.

Uh oh!