@complexthings/superpowers-agent 8.2.2 → 8.3.0

package/.agents/templates/AGENTS.md.template

@@ -7,23 +7,47 @@ Skills extend your capabilities with proven workflows.
 **Before starting any task**, check for relevant skills:
 1. Use your native skill tool to list available skills
 2. Review skill frontmatter/descriptions to identify relevant ones
-3. `superpowers-agent find-skills [PATTERN]` for superpowers skills
+3. Run `superpowers-agent find-skills [PATTERN]` for superpowers skills
 
-### Load Just-In-Time
+### Skill Loading Rules
 
-Only read the full skill content when you're ready to execute it:
-1. **Load fully** when about to use (not before)
-2. **Follow exactly** as written
-3. **Announce:** "Using Skill: {Name} to {Purpose}"
+- Load skills **JIT only** — never preload to "understand" them
+- Follow skill instructions **exactly as written** — no skimming, no shortcuts
+- If a skill has a checklist, create a todo for **each item** — no mental tracking
+- Simple tasks benefit from skills as much as complex ones
 
-{{TOOL_MAPPINGS}}
+**Skill priority (highest to lowest):** Project → Personal → Superpowers
 
-### Version Monitoring
+### Using Tools with Skills
 
-**Bootstrapped:** ^^SAV:{{VERSION}}^^
+1. **Native Skill Tool** — Use your platform's native skill tool to load skills.
+2. **Symlinked Skills** — Superpowers skills are symlinked to your platform's skills directory.
+3. **Fallback** — If a skill isn't found via native tool, run `superpowers-agent find-skills [PATTERN]`
+4. **Tool Equivalence** — When a skill references a tool you don't have, substitute your equivalent:
 
-If superpowers-agent commands show a different version, inform user:
-```
+| Skill References | Use Your Equivalent |
+|---|---|
+| File operations | Read / Write / Edit tools |
+| Terminal commands | Bash / Shell tool |
+| Search operations | Grep / Glob / Search tools |
+| Task management | TodoWrite / TodoRead tools |
+| Subagent dispatch | Task / Agent tool — or execute directly |
+| Web fetching | WebFetch / Fetch tool |
+
+### Version Check
+
+**Bootstrapped Version:** `^^SAV:{{VERSION}}^^`
+
+At conversation start:
+1. Run a superpowers-agent command and note the version string `^^SAV:X.Y.Z^^` in output
+2. Compare against bootstrapped version `^^SAV:{{VERSION}}^^`
+3. If they differ, notify the user:
+   Your superpowers-agent may have updates. Run:
+   ```sh
+   superpowers-agent update && superpowers-agent bootstrap && superpowers-agent setup-skills
+   ```
+
+```sh
 superpowers-agent update && superpowers-agent bootstrap && superpowers-agent setup-skills
 ```
 
@@ -33,8 +57,4 @@ Skills encode proven techniques that prevent mistakes. Not using them means repe
 
 **If a skill exists for your task, you MUST use it.**
 
----
-
-**Detailed reference for `superpowers-agent` if needed by agents:** `.agents/docs/SUPERPOWERS.md`
-
 *Generated by Superpowers on {{DATE}}*

package/.agents/templates/SUPERPOWERS.md.template

@@ -1,9 +1,11 @@
 # SUPERPOWERS Reference Guide
 
+You are an autonomous agent with access to the `superpowers-agent` system. Reference documentation is in the `AGENTS.md` and the detailed reference guide below.
+
 > This is detailed reference documentation for the superpowers-agent system.
 > Load this file only when you need specific information not covered in AGENTS.md.
 
-**Current Version:** ^^SAV:{{VERSION}}^^
+**Bootstrapped Version:** `^^SAV:{{VERSION}}^^`
 
 ---
 
@@ -17,92 +19,24 @@ npm install -g @complexthings/superpowers-agent
 
 ---
 
-## Version Detection Details
-
-When superpowers-agent runs, it displays its version as `^^SAV:X.Y.Z^^` in output.
+## Version Check
 
-**How to check for updates:**
-1. Note the bootstrapped version in AGENTS.md: `^^SAV:{{VERSION}}^^`
-2. Compare with version shown when running superpowers-agent commands
-3. If versions differ, inform user:
-   ```
+At conversation start:
+1. Run a superpowers-agent command and note the version string `^^SAV:X.Y.Z^^` in output
+2. Compare against bootstrapped version `^^SAV:{{VERSION}}^^`
+3. If they differ, notify the user:
    Your superpowers-agent may have updates. Run:
+   ```sh
    superpowers-agent update && superpowers-agent bootstrap && superpowers-agent setup-skills
    ```
 
-**When to check:**
-- At conversation start
-- When commands behave unexpectedly
-- When user asks about updates or capabilities
-
----
-
-## Skill Locations
-
-### Project Skills (highest priority)
-- `.agents/skills/`
-- `.claude/skills/`
-- `.copilot/skills/`
-- `.opencode/skill/`
-- `.cursor/skills/`
-- `.gemini/skills/`
-- `.codex/skills/`
-
-### Personal Skills
-- `~/.agents/skills/`
-- `~/.claude/skills/`
-- `~/.copilot/skills/`
-- `~/.config/opencode/skill/`
-- `~/.cursor/skills/`
-- `~/.gemini/skills/`
-- `~/.codex/skills/`
-
-### Superpowers Skills
-- `~/.agents/superpowers/skills/`
-
-**Priority:** Project > Personal > Superpowers (when names match)
-
----
-
-## Skill Naming Conventions
-
-| Prefix | Source |
-|--------|--------|
-| `skill-name` | Project skills |
-| `claude:skill-name` | Claude skills |
-| `copilot:skill-name` | Copilot skills |
-| `opencode:skill-name` | OpenCode skills |
-| `cursor:skill-name` | Cursor skills |
-| `gemini:skill-name` | Gemini skills |
-| `codex:skill-name` | Codex skills |
-| `superpowers:skill-name` | Superpowers skills |
-
 ---
 
-## Creating New Skills
-
-1. Use the `brainstorming` skill to plan the skill
-2. Use the `writing-skills` skill to create it
-3. Test with `testing-skills-with-subagents`
-4. Save to `.agents/skills/` (project) or `~/.agents/superpowers/skills/` (global)
-
----
-
-## Skills with Checklists
-
-If a skill has a checklist, create todos for EACH item. Mental tracking = steps get skipped.
+## Skill Loading Rules
 
----
-
-## Common Mistakes to Avoid
-
-Reject these rationalizations:
-
-- "I'll read all the skills upfront to understand them" -> Load JIT only
-- "This is too simple for a skill" -> Simple tasks benefit most from proven process
-- "I already know how to do this" -> Skills encode edge cases you'll miss
-- "I'll just skim the skill" -> Follow instructions exactly as written
-
----
+- Load skills **JIT only** — never preload to "understand" them
+- Follow skill instructions **exactly as written** — no skimming, no shortcuts
+- If a skill has a checklist, create a todo for **each item** — no mental tracking
+- Simple tasks benefit from skills as much as complex ones
 
-*Generated by Superpowers on {{DATE}}*
+**Skill priority (highest to lowest):** Project → Personal → Superpowers

package/.github/workflows/main.yaml

@@ -7,6 +7,7 @@ jobs:
   publish:
     runs-on: ubuntu-latest
     permissions:
+      id-token: write
       contents: read
     strategy:
       matrix:

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@complexthings/superpowers-agent",
-  "version": "8.2.2",
+  "version": "8.3.0",
   "description": "Superpowers agent CLI — skills system for AI coding agents",
   "type": "module",
   "bin": {

package/.agents/prompts/001-copilot-tool-mapping-do/001-copilot-tool-mapping-do.md

@@ -1,116 +0,0 @@
----
-number: 001
-topic: copilot-tool-mapping
-purpose: do
-dependencies: []
-created: 2025-11-25
----
-
-# GitHub Copilot Tool Mapping - Do
-
-<objective>
-Generate a comprehensive Tool Mapping section for GitHub Copilot that maps agent-agnostic tool names (used in Superpowers skills) to GitHub Copilot's actual available tools.
-
-Purpose: Enable GitHub Copilot to execute Superpowers skills by providing clear mappings from generic tool names to Copilot-specific equivalents or alternative approaches.
-
-Output: A complete **Tool Mapping for GitHub Copilot:** markdown section that can be inserted into AGENTS.md or similar documentation files.
-</objective>
-
-<context>
-Reference the existing tool mapping structure from:
-@.agents/prompts/current_prompt.md
-
-The current mapping shows the structure, but you need to verify and complete it with GitHub Copilot's actual available tools. The existing mapping is:
-
-```markdown
-**Tool Mapping for GitHub Copilot:**
-When skills reference tools you don't have, substitute your equivalent tools:
-- `TodoWrite` → `manage_todo_list` (your planning/task tracking tool)
-- `Task` tool with subagents → Execute the work yourself, as subagent dispatch isn't available
-- `Skill` tool → `superpowers-agent execute` command (already available)
-- `Read` → `Read File` tool (read file contents)
-- `Write` → `WriteFile` tool (create/overwrite files)
-- `Edit` → `Edit & Reapply` tool (suggest edits to files)
-- `Bash` → `Terminal` tool (execute terminal commands)
-- `List` → `List Directory` tool (read directory structure)
-- `Grep` → `SearchText` tool (search file contents)
-- `Glob` → `FindFiles` tool (find files by pattern)
-- `WebFetch` → `Web` tool (fetch and search web content)
-```
-
-GitHub Copilot context: You are running as GitHub Copilot with access to your native tool set. Examine your available tools and create accurate mappings.
-</context>
-
-<requirements>
-- Include mappings for ALL standard agent-agnostic tools: TodoWrite, TodoRead, Task, Skill, Read, Write, Edit, Bash, List, Grep, Glob, WebFetch, CodebaseSearch, NotebookEdit, ReadLints, DeleteFile
-- Use GitHub Copilot's ACTUAL tool names (not hypothetical or guessed names)
-- Include brief descriptions in parentheses explaining what each tool does
-- For tools GitHub Copilot doesn't have, provide alternative approaches (e.g., "Execute the work yourself")
-- Follow the exact formatting pattern shown in current_prompt.md
-- Be honest about capabilities: if GitHub Copilot lacks a feature, say so explicitly
-- Maintain consistency with other platform mappings (Cursor, Claude Code, Gemini, OpenCode)
-</requirements>
-
-<implementation>
-**Approach:**
-1. List your available tools as GitHub Copilot
-2. Map each agent-agnostic tool name to the closest GitHub Copilot equivalent
-3. For missing capabilities, provide practical workarounds or acknowledge limitations
-4. Format as a markdown bullet list under `**Tool Mapping for GitHub Copilot:**` heading
-
-**What to avoid and WHY:**
-- Don't guess tool names - Use actual GitHub Copilot tool names only
-- Don't omit tools - Include all agent-agnostic tools even if no direct equivalent exists
-- Don't be vague - Provide specific tool names and descriptions
-- Don't copy other platform mappings blindly - Each platform has unique tools
-
-**Integration points:**
-- This mapping will be inserted into AGENTS.md.template
-- Skills will reference these tool names, so accuracy is critical
-- Users depend on these mappings to execute workflows correctly
-</implementation>
-
-<output>
-Create/modify files:
-- `./.agents/prompts/001-copilot-tool-mapping-do/copilot-tool-mapping-do-output.md` - Complete tool mapping section ready to be inserted into AGENTS.md
-
-The output file should contain:
-```markdown
-**Tool Mapping for GitHub Copilot:**
-When skills reference tools you don't have, substitute your equivalent tools:
-- `TodoWrite` → [your actual tool or approach]
-- `TodoRead` → [your actual tool or approach]
-- `Task` tool with subagents → [your actual tool or approach]
-- `Skill` tool → `superpowers-agent execute` command (already available)
-- `Read` → [your actual tool name and description]
-- `Write` → [your actual tool name and description]
-- `Edit` → [your actual tool name and description]
-- `Bash` → [your actual tool name and description]
-- `List` → [your actual tool name and description]
-- `Grep` → [your actual tool name and description]
-- `Glob` → [your actual tool name and description]
-- `WebFetch` → [your actual tool name and description]
-- `CodebaseSearch` → [your actual tool name and description if available]
-- `NotebookEdit` → [your actual tool name and description if available]
-- `ReadLints` → [your actual tool name and description if available]
-- `DeleteFile` → [your actual tool name and description if available]
-```
-</output>
-
-<verification>
-Before declaring complete:
-- Verify all tool names are GitHub Copilot's actual tool names (not guesses)
-- Check that every agent-agnostic tool has a mapping (even if "not available")
-- Confirm descriptions accurately reflect what each tool does
-- Ensure formatting matches the pattern in current_prompt.md
-- Test that a skill using these mappings could be executed by GitHub Copilot
-</verification>
-
-<summary_requirements>
-Create `SUMMARY.md` in same directory with:
-- One-liner: Substantive description of outcome
-- Key Findings: What tools GitHub Copilot has vs doesn't have
-- Decisions Needed: Any ambiguous mappings requiring clarification
-- Blockers: External impediments or "None"
-- Next Step: How to integrate this mapping into AGENTS.md
-</summary_requirements>

package/.agents/prompts/001-copilot-tool-mapping-do/SUMMARY.md

@@ -1,72 +0,0 @@
-# GitHub Copilot Tool Mapping - Summary
-
-**One-liner**: Created comprehensive GitHub Copilot tool mapping with 19 agent-agnostic tools mapped to actual Copilot tool names, including advanced capabilities like subagent dispatch, semantic search, and full Jupyter notebook support.
-
-## Key Findings
-
-### Tools GitHub Copilot Has:
-- ✅ **manage_todo_list** - Complete task management with status tracking
-- ✅ **runSubagent** - Autonomous agent dispatch (different from Claude's multi-turn Task tool)
-- ✅ **read_file** - File reading with chunking support (2000 line limit)
-- ✅ **create_file** - File creation with auto directory creation
-- ✅ **replace_string_in_file / multi_replace_string_in_file** - Exact string replacements
-- ✅ **run_in_terminal** - Full zsh command execution with background process support
-- ✅ **list_dir** - Directory listing
-- ✅ **grep_search** - Text/regex search with file filtering
-- ✅ **file_search** - Glob pattern file finding
-- ✅ **fetch_webpage** - Basic web content extraction
-- ✅ **semantic_search** - Natural language code search
-- ✅ **edit_notebook_file** - Complete Jupyter notebook editing
-- ✅ **run_notebook_cell** - Execute notebook cells
-- ✅ **copilot_getNotebookSummary** - Notebook metadata inspection
-- ✅ **get_errors** - Lint/compile error retrieval
-- ✅ **get_terminal_output** - Background process output retrieval
-- ✅ **get_changed_files** - Git diff support
-
-### Tools GitHub Copilot Doesn't Have:
-- ❌ **Dedicated DeleteFile tool** - Workaround: Use `run_in_terminal` with `rm` command
-- ❌ **AskUserQuestion** - Cannot ask interactive questions during execution (unlike Claude Code)
-- ❌ **Advanced WebFetch** - Has basic content fetch but not Claude's AI-processed web analysis
-- ❌ **MCP Server Integration** - No native MCP support (unlike Cursor)
-
-### Notable Differences:
-- **Subagent Model**: Copilot's `runSubagent` returns single final message vs Claude's conversational Task tool
-- **Todo Management**: Unified `manage_todo_list` vs separated TodoWrite/TodoRead in other platforms
-- **Semantic Search**: Built-in capability that some platforms lack
-- **Notebook Support**: More comprehensive than most platforms
-
-## Decisions Needed
-
-**None** - All mappings are based on actual available tools verified through this session. The mapping is complete and ready for integration.
-
-## Blockers
-
-**None** - All information gathered from active tool availability. No external dependencies required.
-
-## Next Steps
-
-1. **Review and approve** the tool mapping in `copilot-tool-mapping-do-output.md`
-2. **Extract the markdown section** (starting with `**Tool Mapping for GitHub Copilot:**`) from the output file
-3. **Update `.agents/templates/AGENTS.md.template`** with the new comprehensive mapping
-4. **Replace existing mapping** in current AGENTS.md files (if any)
-5. **Test with actual skill execution** to verify mappings work correctly
-6. **Document in RELEASE-NOTES.md** as part of next release
-
-## Validation Checklist
-
-- ✅ All 19 agent-agnostic tools have mappings
-- ✅ All tool names are GitHub Copilot's actual tool names (verified via this session)
-- ✅ Descriptions accurately reflect tool capabilities
-- ✅ Formatting matches pattern from current_prompt.md
-- ✅ Missing capabilities acknowledged with workarounds
-- ✅ Skills using these mappings can be executed by GitHub Copilot
-
-## Additional Value
-
-The output includes:
-- **Tool Capabilities Summary** - Quick reference of what's available/limited
-- **Key Differences section** - Comparison with Claude Code, Cursor, Gemini, OpenCode
-- **Usage Notes** - Important details about tool limitations and best practices
-- **Integration Requirements** - Guidance for incorporating into documentation
-
-This goes beyond a simple mapping to provide actionable context for both skill authors and users.

package/.agents/prompts/001-copilot-tool-mapping-do/copilot-tool-mapping-do-output.md

@@ -1,83 +0,0 @@
-# GitHub Copilot Tool Mapping
-
-## Complete Mapping Section
-
-**Tool Mapping for GitHub Copilot:**
-When skills reference tools you don't have, substitute your equivalent tools:
-- `TodoWrite` → `manage_todo_list` (create and manage task lists with status tracking: not-started, in-progress, completed)
-- `TodoRead` → Check todo list via `manage_todo_list` with `read` operation (view current task states)
-- `Task` tool with subagents → `runSubagent` (dispatch autonomous agents for complex multi-step tasks; agent returns single final message)
-- `Skill` tool → `superpowers-agent execute` command (load and execute skills via CLI)
-- `Read` → `read_file` (read file contents with optional offset/limit for large files; max 2000 lines per call)
-- `Write` → `create_file` (create new files with content; automatically creates parent directories)
-- `Edit` → `replace_string_in_file` or `multi_replace_string_in_file` (exact string replacements; include 3-5 lines context before/after)
-- `Bash` → `run_in_terminal` (execute zsh commands; supports background processes with isBackground flag)
-- `List` → `list_dir` (list directory contents; returns names with trailing / for folders)
-- `Grep` → `grep_search` (search file contents with text or regex patterns; supports includePattern for file filtering)
-- `Glob` → `file_search` (find files by glob pattern; returns matching file paths)
-- `WebFetch` → `fetch_webpage` (fetch and extract main content from web pages for summarization)
-- `CodebaseSearch` → `semantic_search` (natural language search for relevant code in workspace; fallback to `grep_search` for exact matches)
-- `NotebookEdit` → `edit_notebook_file` (edit Jupyter notebooks: insert, edit, or delete cells by cellId)
-- `ReadLints` → `get_errors` (retrieve compile/lint errors for specific files or entire workspace)
-- `DeleteFile` → `run_in_terminal` with `rm` command (no dedicated delete tool; use shell command)
-- `GetTerminalOutput` → `get_terminal_output` (retrieve output from background terminal processes)
-- `Git` → `get_changed_files` (get git diffs of staged/unstaged changes; fallback to `run_in_terminal` for other git operations)
-- `NotebookSummary` → `copilot_getNotebookSummary` (get cell metadata: ids, types, languages, execution info, outputs)
-- `NotebookRun` → `run_notebook_cell` (execute code cells in Jupyter notebooks by cellId)
-
-## Tool Capabilities Summary
-
-### ✅ Available Tools
-- **Task Management**: Full support via `manage_todo_list` and `runSubagent`
-- **File Operations**: Complete read/write/edit capabilities
-- **Terminal**: Full command execution with background process support
-- **Search**: Both semantic and text-based search available
-- **Notebooks**: Complete Jupyter notebook support (edit, run, inspect)
-- **Git**: Basic git diff support, full git via terminal
-- **Diagnostics**: Native lint/error retrieval
-- **Web**: Basic webpage content fetching
-
-### ⚠️ Limited Capabilities
-- **DeleteFile**: No dedicated tool; must use `rm` via terminal
-- **WebFetch**: Basic content extraction only; no AI processing like Claude Code's WebFetch
-
-### 📝 Key Differences from Other Platforms
-
-**vs Claude Code:**
-- GitHub Copilot has `runSubagent` (single-response agents) vs Claude's rich `Task` tool (multi-turn agents)
-- Copilot lacks `AskUserQuestion` (interactive questions during execution)
-- Copilot uses `manage_todo_list` vs Claude's `TodoWrite`/`TodoRead` separation
-- Both have similar file and search capabilities
-
-**vs Cursor:**
-- Copilot has semantic search; Cursor has `codebase_search`
-- Both have similar terminal and file operation tools
-- Cursor has more granular todo operations (`todo_write` vs combined `manage_todo_list`)
-- Cursor has MCP server integration
-
-**vs Gemini/OpenCode:**
-- GitHub Copilot has more advanced notebook support
-- Copilot has semantic search (Gemini/OpenCode may not)
-- Similar basic file/terminal capabilities across all platforms
-
-## Usage Notes
-
-1. **Subagent Limitations**: `runSubagent` returns only a final message, not a conversation. Design tasks for autonomous completion.
-
-2. **Edit Tool Context**: `replace_string_in_file` requires exact string matching with 3-5 lines of context before/after. Use `multi_replace_string_in_file` for efficiency when making multiple edits.
-
-3. **File Reading**: `read_file` truncates at 2000 lines. For larger files, use `offset` and `limit` parameters to read in chunks.
-
-4. **Terminal Background Processes**: Set `isBackground=true` for long-running processes (servers, watch tasks). Retrieve output later with `get_terminal_output`.
-
-5. **Notebook Cells**: Always call `copilot_getNotebookSummary` first to get cellIds before editing or running cells. Don't reference cellIds in user messages (use cell numbers instead).
-
-6. **Search Strategy**: Use `semantic_search` for finding code by meaning/concept. Use `grep_search` for exact string/regex matches.
-
-## Integration Requirements
-
-When inserting this mapping into AGENTS.md or similar files:
-- Place in the appropriate platform section
-- Ensure consistent formatting with other platform mappings
-- Update if GitHub Copilot's tool set changes
-- Test with actual skill execution to verify accuracy

package/.agents/prompts/001-release-management-do/001-release-management-do.md

@@ -1,161 +0,0 @@
----
-number: 001
-topic: release-management
-purpose: do
-dependencies: []
-created: 2025-11-29
----
-
-# Release Management - Do
-
-<objective>
-Create and publish a new version release for the superpowers-agent project
-
-Purpose: Systematically document changes, tag releases, and publish updates to maintain clear version history and enable users to track improvements
-Output: Updated README.md, RELEASE-NOTES.md, git tag, and pushed branches to origin
-</objective>
-
-<context>
-Project structure:
-- `/Users/greg/sites/ai/superpowers/` - Repository root
-- `/Users/greg/sites/ai/superpowers/.agents/` - Agents configuration and superpowers-agent source
-- `README.md` - Main documentation with "What's New" section
-- `RELEASE-NOTES.md` - Comprehensive release history
-- `.agents/package.json` - Version number (authoritative source)
-- Working branch: `auto-updating` (where releases are prepared)
-- Main branch: `main` (where releases are finalized)
-
-@README.md
-@RELEASE-NOTES.md
-@.agents/package.json
-</context>
-
-<requirements>
-- Review git history since last version tag to understand changes
-- Document all changes comprehensively in RELEASE-NOTES.md
-- Update README.md "What's New" section with concise highlights
-- Version number must match semantic versioning (MAJOR.MINOR.PATCH)
-- MAJOR version for breaking changes or complete rewrites
-- MINOR version for new features (backward compatible)
-- PATCH version for bug fixes and documentation updates
-- Create annotated git tag with descriptive message
-- Push changes to auto-updating branch first
-- Merge auto-updating into main
-- Push all branches and tags to origin remote
-- Maintain chronological order in RELEASE-NOTES.md (newest first)
-</requirements>
-
-<implementation>
-Follow this sequence:
-
-1. **Review Changes**
-   - Check current branch: `git branch --show-current`
-   - Review commits since last tag: `git log v{LAST_VERSION}..HEAD --oneline`
-   - Review file changes: `git diff v{LAST_VERSION}..HEAD --stat`
-   - Understand the scope and impact of changes
-
-2. **Update Documentation**
-   - Update README.md "What's New" section:
-     - Add new version at top with date
-     - Include 3-5 bullet points with key highlights
-     - Use emojis for visual hierarchy (🏗️ 📦 🔧 ✅)
-     - Move previous "What's New" to "Previous Releases" section
-   - Update RELEASE-NOTES.md:
-     - Add comprehensive new version section at top (after intro)
-     - Include detailed subsections as appropriate:
-       - Major features/changes
-       - New capabilities
-       - Improvements
-       - Bug fixes
-       - Breaking changes (if any)
-       - Migration guidance
-       - Technical details
-     - Use markdown formatting for clarity
-     - Include code examples where helpful
-
-3. **Commit Documentation**
-   - Stage changes: `git add README.md RELEASE-NOTES.md`
-   - Commit with descriptive message:
-     ```
-     docs: Release v{VERSION} - {Brief Description}
-     
-     - Update README.md with v{VERSION} highlights
-     - Add comprehensive v{VERSION} release notes
-     - {Other notable documentation changes}
-     ```
-
-4. **Create and Push Tag**
-   - Create annotated tag:
-     ```
-     git tag -a v{VERSION} -m "v{VERSION} - {Brief Description}
-     
-     {2-3 sentence summary of release}
-     
-     See RELEASE-NOTES.md for full details."
-     ```
-   - Push tag: `git push origin v{VERSION}`
-
-5. **Push Auto-Updating Branch**
-   - Verify on auto-updating branch
-   - Push: `git push origin auto-updating`
-
-6. **Merge to Main**
-   - Checkout main: `git checkout main`
-   - Merge auto-updating: `git merge auto-updating`
-   - Verify merge successful
-   - Push main: `git push origin main`
-
-Avoid:
-- Generic commit messages ("Update docs", "Version bump")
-- Missing version sections in either README or RELEASE-NOTES
-- Pushing tags before pushing commits
-- Skipping the auto-updating → main merge flow
-</implementation>
-
-<output>
-Modify files:
-- `README.md` - Update "What's New" section with new version highlights
-- `RELEASE-NOTES.md` - Add comprehensive new version section at top
-
-Git operations:
-- Commit documentation updates
-- Create annotated tag: `v{VERSION}`
-- Push tag to origin
-- Push auto-updating branch to origin
-- Merge auto-updating into main
-- Push main branch to origin
-
-Terminal output should confirm:
-```
-[auto-updating {hash}] docs: Release v{VERSION} - {Description}
-To github.com:complexthings/superpowers.git
- * [new tag]         v{VERSION} -> v{VERSION}
-To github.com:complexthings/superpowers.git
-   {old}..{new}  auto-updating -> auto-updating
-Updating {old}..{new}
-Fast-forward
-To github.com:complexthings/superpowers.git
-   {old}..{new}  main -> main
-```
-</output>
-
-<verification>
-Before declaring complete:
-- Verify README.md has new version section at top
-- Verify RELEASE-NOTES.md has comprehensive new version section
-- Verify tag created: `git tag | grep v{VERSION}`
-- Verify tag pushed: `git ls-remote --tags origin | grep v{VERSION}`
-- Verify auto-updating pushed: `git log origin/auto-updating -1`
-- Verify main updated: `git log origin/main -1`
-- Check GitHub repository shows new tag
-- Check that both branches are at same commit
-</verification>
-
-<summary_requirements>
-Create `SUMMARY.md` in same directory with:
-- One-liner: Released v{VERSION} with {key changes}
-- Key Findings: What changed, why it matters, impact on users
-- Decisions Needed: None (or any follow-up releases needed)
-- Blockers: None (or any issues discovered)
-- Next Step: Monitor for issues, plan next release, or update documentation
-</summary_requirements>

package/.agents/prompts/001-release-management-do/SUMMARY.md

@@ -1,19 +0,0 @@
-# Release Management Summary
-
-**One-liner:** Released v6.0.0 documenting complete codebase modernization (3,406-line monolith → 324-line modular executable), created git tag, merged to main, and published reusable release management prompt for future versions.
-
-## Key Findings
-- v6.0.0 represents complete architectural transformation with 22 organized modules across 6 logical layers (CLI, Commands, Core, Integrations, Skills, Utils)
-- 90% bundle reduction achieved through esbuild optimization while maintaining zero breaking changes for users
-- Release process successfully documented in reusable prompt (001-release-management-do.md) for consistent future releases
-- All git operations completed successfully: tag created, branches pushed, auto-updating merged to main
-- README.md and RELEASE-NOTES.md comprehensively updated with modular architecture details
-
-## Decisions Needed
-None
-
-## Blockers
-None
-
-## Next Step
-Monitor repository for any issues with v6.0.0 release and use the newly created release management prompt (001-release-management-do.md) for future version releases.