feat(tools): add summarize and bird CLI subagents (t034, t035)

.agent/AGENTS.md

@@ -351,6 +351,8 @@ Subagents provide specialized capabilities. Read them when tasks require domain
 | `memory/` | Cross-session memory - SQLite FTS5 storage, /remember and /recall commands | README (system docs) |
 | `seo/` | Search optimization - keyword research, rankings, site audits, E-E-A-T scoring | dataforseo, serper, google-search-console, site-crawler, eeat-score, domain-research |
 | `content/` | Content creation - copywriting standards, editorial guidelines, tone of voice | guidelines |
+| `tools/content/` | Content tools - summarization, extraction, processing | summarize |
+| `tools/social-media/` | Social media tools - X/Twitter CLI, posting, reading | bird |
 | `tools/build-agent/` | Agent design - composing efficient agents, reviewing agent instructions | build-agent, agent-review |
 | `tools/build-mcp/` | MCP development - creating Model Context Protocol servers and tools | build-mcp, api-wrapper, server-patterns, transports, deployment |
 | `tools/ai-assistants/` | AI tool integration - configuring assistants, CAPTCHA solving, multi-modal agents | agno, capsolver, windsurf, configuration, status |
@@ -434,6 +436,8 @@ For AI-assisted setup guidance, see `aidevops/setup.md`.
 | Agent design | `tools/build-agent/build-agent.md`, `tools/build-agent/agent-review.md` |
 | Database migrations | `workflows/sql-migrations.md` |
 | Framework internals | `aidevops/architecture.md` (when working in `~/Git/aidevops/`) |
+| Content summarization | `tools/content/summarize.md` |
+| X/Twitter automation | `tools/social-media/bird.md` |
 
 ## Security
 

.agent/tools/content/summarize.md

@@ -0,0 +1,353 @@
+---
+description: URL/YouTube/podcast summarization using steipete/summarize CLI
+mode: subagent
+tools:
+  read: true
+  write: false
+  edit: false
+  bash: true
+  glob: true
+  grep: true
+  webfetch: true
+  task: true
+---
+
+# Summarize CLI Integration
+
+<!-- AI-CONTEXT-START -->
+
+## Quick Reference
+
+- **Purpose**: Summarize URLs, YouTube videos, podcasts, and files using AI
+- **Install**: `npm i -g @steipete/summarize` or `brew install steipete/tap/summarize`
+- **Repo**: https://github.com/steipete/summarize (726+ stars)
+- **Website**: https://summarize.sh
+
+**Quick Commands**:
+
+```bash
+# Summarize any URL
+summarize "https://example.com"
+
+# YouTube video
+summarize "https://youtu.be/dQw4w9WgXcQ" --youtube auto
+
+# Podcast RSS feed
+summarize "https://feeds.npr.org/500005/podcast.xml"
+
+# Local file (PDF, images, audio/video)
+summarize "/path/to/file.pdf" --model google/gemini-3-flash-preview
+
+# Extract content only (no summary)
+summarize "https://example.com" --extract --format md
+```
+
+**Key Features**:
+
+- URLs, files, and media: web pages, PDFs, images, audio/video, YouTube, podcasts, RSS
+- Real extraction pipeline: fetch -> clean -> Markdown (readability + markitdown)
+- Transcript-first media flow: published transcripts when available, Whisper fallback
+- Streaming TTY output with Markdown rendering
+- Local, paid, and free models: OpenAI-compatible local endpoints, paid providers, OpenRouter free preset
+
+**Env Vars**: `OPENAI_API_KEY`, `ANTHROPIC_API_KEY`, `GEMINI_API_KEY`, `XAI_API_KEY`, `OPENROUTER_API_KEY`
+
+<!-- AI-CONTEXT-END -->
+
+## Installation
+
+### npm (recommended)
+
+```bash
+# Global install
+npm i -g @steipete/summarize
+
+# One-shot (no install)
+npx -y @steipete/summarize "https://example.com"
+```
+
+### Homebrew (macOS Apple Silicon)
+
+```bash
+brew install steipete/tap/summarize
+```
+
+### Requirements
+
+- Node.js 22+
+- Optional: `yt-dlp` for YouTube audio extraction
+- Optional: `whisper.cpp` for local transcription
+- Optional: `uvx markitdown` for enhanced preprocessing
+
+## Usage
+
+### Basic Summarization
+
+```bash
+# Web page
+summarize "https://example.com"
+
+# With specific model
+summarize "https://example.com" --model openai/gpt-5-mini
+
+# Auto model selection (default)
+summarize "https://example.com" --model auto
+```
+
+### YouTube Videos
+
+```bash
+# Auto-detect transcript method
+summarize "https://youtu.be/dQw4w9WgXcQ" --youtube auto
+
+# Supports youtube.com and youtu.be URLs
+summarize "https://www.youtube.com/watch?v=dQw4w9WgXcQ"
+```
+
+### Podcasts
+
+```bash
+# RSS feed (transcribes latest episode)
+summarize "https://feeds.npr.org/500005/podcast.xml"
+
+# Apple Podcasts episode
+summarize "https://podcasts.apple.com/us/podcast/2424-jelly-roll/id360084272?i=1000740717432"
+
+# Spotify episode (best-effort)
+summarize "https://open.spotify.com/episode/5auotqWAXhhKyb9ymCuBJY"
+```
+
+### Local Files
+
+```bash
+# PDF (Google models work best)
+summarize "/path/to/document.pdf" --model google/gemini-3-flash-preview
+
+# Images
+summarize "/path/to/image.png"
+
+# Audio/Video
+summarize "/path/to/video.mp4" --video-mode transcript
+```
+
+### Output Control
+
+```bash
+# Length presets: short, medium, long, xl, xxl
+summarize "https://example.com" --length long
+
+# Character target
+summarize "https://example.com" --length 20k
+
+# Hard token cap
+summarize "https://example.com" --max-output-tokens 2000
+
+# Output language
+summarize "https://example.com" --language en
+summarize "https://example.com" --lang auto  # match source
+```
+
+### Content Extraction
+
+```bash
+# Extract content without summarizing
+summarize "https://example.com" --extract
+
+# Extract as Markdown
+summarize "https://example.com" --extract --format md
+
+# Extract as plain text
+summarize "https://example.com" --extract --format text
+```
+
+### Output Formats
+
+```bash
+# JSON output with diagnostics
+summarize "https://example.com" --json
+
+# Plain output (no ANSI/colors)
+summarize "https://example.com" --plain
+
+# Disable streaming
+summarize "https://example.com" --stream off
+```
+
+## Model Configuration
+
+### Supported Providers
+
+| Provider | Model ID Format | API Key |
+|----------|-----------------|---------|
+| OpenAI | `openai/gpt-5-mini` | `OPENAI_API_KEY` |
+| Anthropic | `anthropic/claude-sonnet-4-5` | `ANTHROPIC_API_KEY` |
+| Google | `google/gemini-3-flash-preview` | `GEMINI_API_KEY` |
+| xAI | `xai/grok-4-fast-non-reasoning` | `XAI_API_KEY` |
+| Z.AI | `zai/glm-4.7` | `Z_AI_API_KEY` |
+| OpenRouter | `openrouter/openai/gpt-5-mini` | `OPENROUTER_API_KEY` |
+
+### Free Models via OpenRouter
+
+```bash
+# Setup free model preset
+OPENROUTER_API_KEY=sk-or-... summarize refresh-free --set-default
+
+# Use free models
+summarize "https://example.com" --model free
+```
+
+### Configuration File
+
+Location: `~/.summarize/config.json`
+
+```json
+{
+  "model": { "id": "openai/gpt-5-mini" }
+}
+```
+
+Or shorthand:
+
+```json
+{
+  "model": "openai/gpt-5-mini"
+}
+```
+
+## Chrome Extension
+
+Summarize includes a Chrome Side Panel extension for one-click summarization.
+
+### Setup
+
+1. Install CLI: `npm i -g @steipete/summarize`
+2. Build extension: `pnpm -C apps/chrome-extension build`
+3. Load in Chrome: `chrome://extensions` -> Developer mode -> Load unpacked
+4. Pick: `apps/chrome-extension/.output/chrome-mv3`
+5. Open Side Panel -> copy install command
+6. Run: `summarize daemon install --token <TOKEN>`
+
+### Daemon Commands
+
+```bash
+# Check daemon status
+summarize daemon status
+
+# Restart daemon
+summarize daemon restart
+```
+
+## Advanced Features
+
+### Firecrawl Fallback
+
+For blocked or thin content, Firecrawl can be used as fallback:
+
+```bash
+# Auto fallback (default)
+summarize "https://example.com" --firecrawl auto
+
+# Force Firecrawl
+summarize "https://example.com" --firecrawl always
+
+# Disable Firecrawl
+summarize "https://example.com" --firecrawl off
+```
+
+Requires `FIRECRAWL_API_KEY` environment variable.
+
+### Whisper Transcription
+
+For audio/video without transcripts:
+
+```bash
+# Force transcription mode
+summarize "/path/to/video.mp4" --video-mode transcript
+
+# Environment variables
+export SUMMARIZE_WHISPER_CPP_MODEL_PATH=/path/to/model.bin
+export SUMMARIZE_WHISPER_CPP_BINARY=whisper-cli
+export SUMMARIZE_DISABLE_LOCAL_WHISPER_CPP=1  # force remote
+```
+
+### Markdown Conversion
+
+```bash
+# Readability mode (default)
+summarize "https://example.com" --markdown-mode readability
+
+# LLM conversion
+summarize "https://example.com" --markdown-mode llm
+
+# Auto (LLM when configured)
+summarize "https://example.com" --markdown-mode auto
+
+# Disable
+summarize "https://example.com" --markdown-mode off
+```
+
+## Common Flags
+
+| Flag | Description |
+|------|-------------|
+| `--model <provider/model>` | Model to use (default: `auto`) |
+| `--timeout <duration>` | Request timeout (`30s`, `2m`, `5000ms`) |
+| `--retries <count>` | LLM retry attempts (default: 1) |
+| `--length <preset\|chars>` | Output length control |
+| `--language, --lang` | Output language (`auto` = match source) |
+| `--max-output-tokens` | Hard cap for LLM output tokens |
+| `--stream auto\|on\|off` | Stream LLM output |
+| `--plain` | No ANSI/OSC Markdown rendering |
+| `--no-color` | Disable ANSI colors |
+| `--format md\|text` | Content format |
+| `--extract` | Print extracted content and exit |
+| `--json` | Machine-readable output |
+| `--verbose` | Debug/diagnostics on stderr |
+| `--metrics off\|on\|detailed` | Metrics output |
+
+## Integration with aidevops
+
+### Use Cases
+
+1. **Research**: Summarize articles, papers, documentation
+2. **Content Curation**: Extract key points from multiple sources
+3. **Podcast Notes**: Generate show notes from episodes
+4. **Video Summaries**: Create text summaries of YouTube content
+5. **Document Processing**: Summarize PDFs and reports
+
+### Example Workflow
+
+```bash
+# Research a topic
+summarize "https://docs.example.com/api" --length long --json > api-summary.json
+
+# Summarize multiple sources
+for url in "${urls[@]}"; do
+  summarize "$url" --length medium >> research-notes.md
+done
+
+# Extract and process
+summarize "https://example.com" --extract --format md | process-content.sh
+```
+
+## Troubleshooting
+
+### Common Issues
+
+1. **Model not responding**: Check API key is set correctly
+2. **YouTube transcript unavailable**: Install `yt-dlp` for audio extraction
+3. **PDF extraction failing**: Use Google models for best PDF support
+4. **Rate limiting**: Add `--timeout` and `--retries` flags
+
+### Debug Mode
+
+```bash
+summarize "https://example.com" --verbose
+```
+
+## Resources
+
+- **GitHub**: https://github.com/steipete/summarize
+- **Website**: https://summarize.sh
+- **npm**: https://www.npmjs.com/package/@steipete/summarize
+- **Docs**: https://github.com/steipete/summarize/tree/main/docs