Files
2026-04-20 22:51:41 +08:00

381 lines
14 KiB
Markdown

# Changelog
All notable changes to OPC Skills are documented here.
The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
## Skill Compatibility & Dependency Matrix
Each skill maintains its own independent version. Use this matrix to understand dependencies and compatibility.
| Skill | Current Version | Requires | Min Versions |
|-------|-----------------|----------|--------------|
| **requesthunt** | 2.3.0 | - | - |
| **domain-hunter** | 1.0.0 | twitter, reddit | twitter ≥1.0.0, reddit ≥1.0.0 |
| **logo-creator** | 1.0.0 | nanobanana | nanobanana ≥1.0.0 |
| **banner-creator** | 1.0.0 | nanobanana | nanobanana ≥1.0.0 |
| **nanobanana** | 1.0.0 | - | - |
| **reddit** | 1.0.0 | - | - |
| **twitter** | 1.0.0 | - | - |
| **producthunt** | 1.0.0 | - | - |
| **seo-geo** | 1.0.0 | twitter, reddit | twitter ≥1.0.0, reddit ≥1.0.0 |
| **archive** | 1.1.0 | - | - |
**Key Points:**
- Each skill has an independent version number (MAJOR.MINOR.PATCH)
- Skills can be released independently without coordinating with others
- Use this matrix to identify dependencies before updating a skill
- When a dependency skill updates with breaking changes, dependent skills need to be tested
---
## [Unreleased]
## Released Versions
## [1.3.0] - 2026-04-20
### requesthunt
- **Version**: 2.2.1 → 2.3.0
- Added Amazon as a new supported data source across all commands (search, list, scrape)
- Added Amazon to platform strengths table, recommended platforms by category, and quick selection rules
- Added Consumer Electronics as a new category recommendation
- Updated all example commands to include `amazon` in platform lists
- Noted Amazon depth cap (max 5) in cost section
- Corresponds to requesthunt v1.14.0 release
## [1.2.1] - 2026-04-19
### requesthunt
- **Version**: 2.2.0 → 2.2.1
- **Security**: Documented SHA256 checksum verification and GitHub Releases source for CLI installer
- **Security**: Added build-from-source alternative (`cargo install --path cli`)
- **Security**: Recommend environment variable (`REQUESTHUNT_API_KEY`) over plaintext command-line API key
- **Security**: Added Content Safety section with untrusted-input handling guidelines for agents
- Addresses security audit findings from Agent Trust Hub (FAIL), Socket (WARN), and Snyk (FAIL) on skills.sh
## [1.2.0] - 2026-04-15
### requesthunt
- **Version**: 2.0.0 → 2.1.0
- Added YouTube and LinkedIn as supported data sources across all commands (search, list, scrape)
- Updated skill description, platform references, and command examples to include 5 platforms
## [1.1.1] - 2026-03-31
### requesthunt
#### [2.0.1] - 2026-03-31
- **Changed**: Updated auth documentation to describe device code flow (visit URL + enter code) instead of generic browser approval
## [1.1.0] - 2026-03-31
### requesthunt
#### [2.0.0] - 2026-03-31
- **Changed**: Switched from Python scripts to `requesthunt` Rust CLI as the sole interface
- **Added**: Browser authentication via `requesthunt auth login` (with `config set-key` fallback for headless/CI)
- **Added**: CLI commands in skills.json registry (search, list, scrape start/status, topics, usage)
- **Added**: Link to agent setup guide (`https://requesthunt.com/setup.md`)
- **Added**: TOON output mode documentation (Token-Oriented Object Notation)
- **Changed**: Switched usage and pricing documentation from cached/realtime quotas to the unified credits model
- **Removed**: Python scripts (`scripts/`) — replaced entirely by CLI commands
- **Fixed**: Updated RequestHunt settings links to use `/dashboard`
## [1.0.11] - 2026-03-13
### archive
#### [1.1.0] - 2026-03-13
- **Added**: Claude Code / Droid plugin support (`.factory-plugin/plugin.json`)
- **Added**: `hooks/hooks.json` — `SessionStart` hook auto-loads `.archive/MEMORY.md` into session context when plugin is installed, enabling cross-session knowledge reuse without manual configuration
- **Added**: `hooks/load-memory.py` — supports `FACTORY_PROJECT_DIR`, `CLAUDE_PROJECT_DIR`, and `cwd()` fallback for cross-platform compatibility
## [1.0.10] - 2026-02-23
### archive
#### [1.0.1] - 2026-02-23
- **Fixed**: Description YAML block scalar (`>-`) replaced with single-line string to fix broken description display on skills.sh and other parsers
## [1.0.9] - 2026-02-23
### Blog
- **Added**: Archive skill announcement blog post (#62)
- "Stop Losing Context Between AI Sessions: Introducing the Archive Skill"
- Includes OG banner image
## [1.0.8] - 2026-02-23
### Skills
- **Added**: Archive skill for indexed session knowledge (#57, #59)
- Archive session learnings, debugging solutions, and deployment logs
- Maintains `.archive/MEMORY.md` index for cross-session knowledge reuse
- Includes logo, install commands, README listing, and website integration
- **Added**: `add-new-opc-skill` checklist skill under `.factory/skills/` (#60)
- Comprehensive guide for adding new skills to the project
### Infrastructure
- **Fixed**: Replace broken `curl -fsSL opc.dev/install.sh` with `npx skills add` (#55, #58)
- install.sh endpoint was returning 404
- Updated all install commands in skills.json and website/worker.js
- Redirect `/install.sh` to GitHub README installation section
## [1.0.7] - 2026-02-15
### Infrastructure
- **Added**: Twitter ([@Yilin0x](https://x.com/Yilin0x)) and Discord badges to README (#52)
- **Changed**: Ranked seo-geo skill at the top of skills list in skills.json and README (#53)
## [1.0.6] - 2026-02-15
### Infrastructure
- **Fixed**: Install stats scraper now handles abbreviated count formats (e.g. "2.8K") from skills.sh (#50)
- Added `_parse_count()` helper for K-suffix abbreviations
- Previously caused seo-geo to steal logo-creator's count and logo-creator to be dropped entirely
### Skills
- **Refactored**: Aligned all SKILL.md frontmatter with Anthropic skill standard (#51)
- Removed `triggers` field from all 9 skills and the template
- Merged trigger keywords into `description` field as "Use when..." clauses
- Frontmatter now only uses `name` and `description` fields
- Fixes seo-geo broken description display on skills.sh (YAML `|` block rendered as literal `|`)
## [1.0.5] - 2026-01-29
### Infrastructure
- **Fixed**: Removed `pluginRoot` and updated all source paths to work around Claude Code bug
- Removed `pluginRoot: "./skills"` from marketplace metadata (Claude Code doesn't respect this field)
- Updated all plugin source paths from `"./domain-hunter"` to `"./skills/domain-hunter"`
- Fixes "Source path does not exist" error when installing plugins
- Workaround for known Claude Code issues #11243 and #11278
### Skills
- (no skill version changes in this release)
## [1.0.4] - 2026-01-29
### Infrastructure
- **Fixed**: Updated marketplace metadata version to trigger cache refresh
- Changed marketplace metadata version from 1.0.0 to 1.0.3
- Fixes "Source path does not exist" error when installing plugins in Claude Code
- Ensures Claude Code refreshes marketplace cache with correct plugin paths
### Skills
- (no skill version changes in this release)
## [1.0.3] - 2026-01-29
### Website
- **Fixed**: Blog OG image path for installation tutorial post
- Corrected image path from `2026-01-28-install-tutorial-og.png` to `2026-01-28-install-opc-skills-claude-code-og.png`
- Fixes broken Open Graph preview on social media shares
### Skills
- (no skill version changes in this release)
## [1.0.2] - 2026-01-29
### Infrastructure
- **Fixed**: Claude Code Plugin Marketplace schema validation errors
- Updated all plugin source paths from bare names to relative paths (e.g., `"requesthunt"` → `"./requesthunt"`)
- Ensures compatibility with Claude Code's `/plugin marketplace add` command
- Resolves "Invalid input" errors when adding marketplace
### Skills
- (no skill version changes in this release)
## [1.0.1] - 2026-01-29
### Website
- **Added**: New blog post "Why Skills Beat Docs: The Rise of Agent-Native Documentation"
- Analysis of 100x engagement gap between docs and skills
- Comprehensive ecosystem overview (Skills.sh, Mintlify, awesome-claude-code)
- Step-by-step guide for converting docs to skills
### Skills
- (no skill version changes in this release)
### requesthunt
#### [1.0.0] - 2025-01-21
- **Added**: Initial release
- Generate user demand research reports from Reddit, X, and GitHub
- Collect real user feedback across multiple platforms
- Filter and search by topic, platform, and timeframe
- Generate structured demand research reports
### domain-hunter
#### [1.0.0] - 2025-01-21
- **Added**: Initial release
- Search domains, compare registrar prices, and find promo codes
- Query domain availability and pricing
- Compare prices across registrars
- Find current promo codes from Twitter and Reddit
- **Dependencies**: twitter ≥1.0.0, reddit ≥1.0.0
### logo-creator
#### [1.0.0] - 2025-01-21
- **Added**: Initial release
- Create logos using AI image generation
- Generate logo variations with Gemini
- Remove background from images
- Crop logos to desired aspect ratios
- Export as SVG vector format
- **Dependencies**: nanobanana ≥1.0.0
### banner-creator
#### [1.0.0] - 2025-01-21
- **Added**: Initial release
- Create banners for GitHub, Twitter, LinkedIn, and other platforms
- Generate banner variations with Gemini
- Crop to platform-specific ratios (16:9, 21:9, 2:1, etc.)
- Support for different banner formats and styles
- **Dependencies**: nanobanana ≥1.0.0
### nanobanana
#### [1.0.0] - 2025-01-21
- **Added**: Initial release
- Generate and edit images using Google Gemini 3 Pro Image (Nano Banana Pro)
- Text-to-image generation
- Image-to-image editing and variations
- Support for multiple aspect ratios (1:1, 2:3, 3:2, 16:9, 21:9, etc.)
- 2K and 4K high-resolution output options
- Batch image generation
### reddit
#### [1.0.0] - 2025-01-21
- **Added**: Initial release
- Search and retrieve content from Reddit
- Access public JSON API without authentication
- Search posts and subreddits
- Get user profiles and comment threads
- No API key required
### twitter
#### [1.0.0] - 2025-01-21
- **Added**: Initial release
- Search and retrieve content from Twitter/X
- User information and tweet retrieval
- Search tweets by keyword
- Get follower information and trends
- Via twitterapi.io API service
### producthunt
#### [1.0.0] - 2025-01-21
- **Added**: Initial release
- Search and retrieve content from Product Hunt
- Query posts, topics, and collections
- Get user and product information
- Access GraphQL API for detailed data
- Requires Product Hunt API access token
### seo-geo
#### [1.0.0] - 2025-01-21
- **Added**: Initial release
- SEO & GEO (Generative Engine Optimization) for websites
- Optimize for traditional search engines (Google, Bing)
- Optimize for AI search engines (ChatGPT, Perplexity, Gemini, Copilot, Claude)
- Generate schema markup and JSON-LD
- Keyword research and SERP analysis
- Princeton GEO research methods for +40% AI visibility
- Optional DataForSEO API integration for advanced features
- **Dependencies**: twitter ≥1.0.0, reddit ≥1.0.0
---
## Initial Release Features
### Unified Installation & Support
- Unified installation system via `npx skills add`
- Support for 16+ AI agent tools (Claude Code, Cursor, Droid, Windsurf, etc.)
- Composable skills with dependency management
- Comprehensive documentation website (opc.dev)
- SKILL.md standard with YAML frontmatter for all skills
### Documentation & Resources
- Official website: https://opc.dev
- Skill browser: https://skills.sh/ReScienceLab/opc-skills
- Individual skill repositories on GitHub
- Example workflows in each skill directory
- API documentation and rate limit information
### Infrastructure
- GitHub repository: https://github.com/ReScienceLab/opc-skills
- MIT License
- Automated skill installation scripts
- Website deployment pipeline
## Version Compatibility
| Version | Status | Release Date | Notable Changes |
|---------|--------|--------------|-----------------|
| 1.1.0 | Stable | 2026-03-31 | requesthunt v2.0.0 — CLI-first, Python scripts removed |
| 1.0.0 | Stable | 2025-01-21 | Initial release with 9 core skills |
## Migration Guides
### Coming Soon
Migration guides for major version upgrades will be documented here.
## Notes
### API Keys Required
- **requesthunt**: CLI device code auth (`requesthunt auth login`) or REQUESTHUNT_API_KEY (requesthunt.com/dashboard)
- **twitter**: TWITTERAPI_API_KEY (twitterapi.io, ~$0.15-0.18/1k requests)
- **logo-creator**: GEMINI_API_KEY, REMOVE_BG_API_KEY, RECRAFT_API_KEY
- **banner-creator**: GEMINI_API_KEY (Google AI Studio)
- **nanobanana**: GEMINI_API_KEY
- **producthunt**: PRODUCTHUNT_ACCESS_TOKEN
- **seo-geo**: DATAFORSEO_LOGIN, DATAFORSEO_PASSWORD (optional)
### Rate Limits
- **requesthunt**: Free tier 100 credits/month at 10 req/min, Pro tier 2,000 credits/month at 60 req/min, 1 credit per API call, scrapes cost depth x platforms
- **twitter**: Depends on twitterapi.io plan
- **nanobanana**: Google Gemini API limits apply
- **seo-geo**: DataForSEO API limits apply
### Supported Platforms
- Claude Code (Desktop)
- Cursor
- Factory Droid
- Windsurf
- OpenCode
- Codex
- GitHub Copilot
- Gemini CLI
- Goose
- Kilo Code
- Roo Code
- Trae
- And more via `npx skills add`
## Contributing
Interested in contributing? Please see:
- Contributing Guidelines: https://github.com/ReScienceLab/opc-skills/blob/main/CONTRIBUTING.md (coming soon)
- Issue Tracker: https://github.com/ReScienceLab/opc-skills/issues
- Skill Template: https://github.com/ReScienceLab/opc-skills/tree/main/template
## Support
For issues and questions:
- GitHub Issues: https://github.com/ReScienceLab/opc-skills/issues
- Website: https://opc.dev
- Documentation: https://skills.sh/ReScienceLab/opc-skills
## License
All OPC Skills are released under the [MIT License](https://github.com/ReScienceLab/opc-skills/blob/main/LICENSE).
---
Generated: 2025-01-21
Project: OPC Skills - AI Agent Skills for Solopreneurs