Add memory-engine and site-automation skills

[adibis-git]

Summary

This PR contributes two production-tested skills from the OpenClaw project to the CommandCodeAI/agent-skills repository:

1. Memory Engine — Persistent, searchable memory across sessions with semantic naming and hybrid search
2. Site Automation — Website content management via automation APIs (projects, changelogs, posts, newsletters, comments)

Both skills have been battle-tested in production and are broadly applicable to any agent managing long-term context or website content.

Skills Included

1. Memory Engine (skills/memory-engine/SKILL.md)

What it does: Stores and retrieves long-term knowledge across sessions using semantic file naming and hybrid search (semantic embeddings + keyword fallback).

Why valuable:

• Solves the "agent amnesia" problem — agents can now remember decisions, context, and past work
• Semantic naming convention makes files discoverable and organized
• Hybrid search works with or without embeddings API
• Directly applicable to any multi-session agent workflow

Unique aspects:

• Semantic file naming: YYYY-MM-DD__[CATEGORY]__[TOPIC]__[CONTEXT].md
• Adaptive search with query type detection
• Diagnostics for debugging search issues
• Graceful fallback from semantic to keyword search

Use cases:

• Recall past decisions before making new ones
• Store summaries of completed work
• Search for context from previous sessions
• Build institutional knowledge over time

2. Site Automation (skills/site-automation/SKILL.md)

What it does: Manages website content through automation APIs — sync GitHub projects, generate changelogs, create posts, send newsletters, manage comments.

Why valuable:

• Pattern for integrating with web applications and content management
• Useful for content creators, developers, and agents managing websites
• Demonstrates clean API wrapper pattern
• Covers full content lifecycle (creation, distribution, engagement)

Unique aspects:

• Unified CLI wrapper for multiple API endpoints
• GitHub integration for automatic changelog generation
• Newsletter management with idempotency
• Comment moderation and response workflow

Use cases:

• Automate blog post publishing
• Generate changelogs from git commits
• Send newsletters to subscribers
• Manage user comments and engagement
• Sync project data from GitHub

Quality Checklist

• Skills follow CommandCodeAI standard format (YAML frontmatter + sections)
• "When to Use This Skill" sections with explicit trigger phrases
• Comprehensive command examples (copy-pasteable)
• Real-world usage examples
• Clear guardrails and safety considerations
• All dependencies documented
• Generalized for broad applicability (not project-specific)
• Tested in production environments

Testing

Both skills have been:

• Used in production by OpenClaw agents
• Tested with multiple agents and workflows
• Validated for clarity and completeness
• Reviewed for security and best practices

Future Contributions

OpenClaw has additional skills ready for contribution in Tier 2:

• Content Moderation — Automated content moderation with prompt injection detection
• Swarm Controller — Multi-agent orchestration for complex research tasks
• BMAD Method Bridge — Structured workflow integration

Notes

• These skills are general-purpose and not project-specific
• They follow the CommandCodeAI design philosophy and patterns
• Both have been refined based on real-world usage
• Documentation is comprehensive and includes guardrails

---

Contributor: OpenClaw Project
Date: February 25, 2026