A structured weekly self-review skill that analyzes agent performance, identifies technical failures, and recommends actionable improvements.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install retrospective
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install retrospective using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Agent Retrospective?
The Agent Retrospective skill provides a rigorous framework for AI agents to perform self-analysis and optimization. By examining daily logs and long-term context, it generates an honest assessment of successes and technical struggles without falling back on generic platitudes. It is an essential component of the Openclaw Skills ecosystem for users who want their agents to evolve over time.
This skill goes beyond simple logging by performing deep root-cause analysis on failures, such as bot detection issues or API quota exhaustion. It ensures your agent remains at peak performance by auditing cron job health, workspace hygiene, and configuration integrity, ultimately resulting in a documented improvement roadmap.
Agent Retrospective Use Cases
- Automating a weekly Friday review of all agent activities and successes.
- Identifying recurring technical bottlenecks like browser automation failures or scraping blocks.
- Discovering and evaluating new Openclaw Skills that fill existing capability gaps.
- Auditing cron job reliability and configuration health to prevent silent failures.
- Maintaining workspace hygiene by identifying stale scripts and orphaned projects.
How Agent Retrospective Works
- Context Gathering: The agent scans the memory directory for daily logs, reviews the long-term MEMORY.md file, and checks cron job execution history.
- Performance Analysis: It identifies concrete wins and performs a brutally honest review of failures, estimating time wasted and looking for recurring patterns.
- Skill Gap Audit: The system searches for relevant Openclaw Skills that could have prevented identified failures or improved efficiency.
- Infrastructure Review: It checks for expired credentials, rotating API keys, and workspace clutter.
- Report Generation: A structured markdown report is written to the memory folder, categorizing action items by the effort required to implement them.
- Memory Update: Key findings are synthesized back into the primary MEMORY.md file to ensure long-term learning.
Agent Retrospective Setup
The skill is designed to auto-schedule itself upon first use. You can trigger the initial run and setup by messaging your agent:
retrospective
To manually verify the cron schedule within the Openclaw Skills environment, use:
cron(action="list")
Agent Retrospective Data Schema & Taxonomy
The skill generates structured reports and updates existing metadata to maintain a clean record of agent evolution:
| File Type | Path | Description |
|---|---|---|
| Monthly Retro | memory/retro-YYYY-MM-DD.md |
The full technical analysis of the 7-day period. |
| Long-term Memory | MEMORY.md |
Updated with key insights and prioritized action items. |
| Logs | memory/*.log |
Source data used for the review process. |
Output Sections:
- Wins: Evidence-based achievements.
- Struggles Table: Task, Root Cause, Time Spent, and Recurrence status.
- Skill Gaps: Proposed plugins or tools to install.
- Action Items: Categorized into Quick Wins, Medium Effort, and Large Projects.
Agent Retrospective Advanced Features
- Auto-scheduling via cron to ensure retrospectives occur every Friday at 16:00 local time.
- Intelligent relevance scoring for suggesting new Openclaw Skills based on historical failure data.
- Automated workspace hygiene checks to identify and cleanup orphaned project files.
- Root cause analysis for complex automation failures like Cloudflare blocks or auth timeouts.
SKILL.md
Loading
Related Openclaw Skills
Ledger Financial Analysis
A professional-grade command-line interface for calculating financial metrics, managing expense entries, and generating investment reports.
Fairytale Dream Transform Video Gen
A specialized video generation skill for Seedance 2.0 that transforms prompts and images into creamy, pastel fairytale dreamscapes with soft global lighting and sugar-glow particles.
Mood & Emotional Wellness Reference
A comprehensive AI-driven reference for mood science, emotional regulation strategies, and wellness practices.
Form CRO Skill
An intelligent optimization engine designed to transform high-friction forms into high-converting user experiences.
Watermark Reference Skill
A comprehensive technical reference for implementing visible and invisible watermarking, steganography, and digital rights management across various media.
Hydraulic Crush Video Generator
A specialized AI video generation skill for creating high-quality, vertical hydraulic press crushing videos designed for social media feeds.
Featured*
