minds-eye

Remember and retrieve visual content from conversations. Use when: (1) user sends an image, photo, chart, diagram, or screenshot and wants it saved/remembere...

This skill appears to do what it says: it analyzes images by sending them to the configured vision-capable model and stores copies and metadata under ~/.multimodal-memory/. Before installing, verify which provider/API key the skill will use (check ~/.openclaw/openclaw.json and your OPENAI_API_KEY). Do not supply file paths to sensitive local files (password screenshots, private keys, etc.), because any image file you point the skill to will be read and uploaded to the API. If you plan to capture websites, note Playwright (or headless Chrome) will load remote pages locally to take screenshots — that can execute page JS. Consider using a separate/limited account or running this skill in an isolated environment if you need to limit exposure, and review or rotate API keys regularly. Finally, inspect ~/.multimodal-memory/ if you want to audit stored images and memory.md summaries or remove them when no longer needed.

Type: OpenClaw Skill Name: minds-eye Version: 1.0.0 The 'minds-eye' skill provides a legitimate mechanism for an OpenClaw agent to store and recall visual information. It includes scripts for capturing website screenshots (capture_url.py), analyzing images via the OpenAI API (analyze.py), and managing a local SQLite database for metadata (store.py, search.py, list.py). The skill correctly retrieves API keys from the standard OpenClaw configuration directory (~/.openclaw/) and maintains a human-readable summary in ~/.multimodal-memory/memory.md. No evidence of data exfiltration, malicious execution, or prompt injection attacks was found; the code is well-structured and aligns perfectly with its stated purpose.