Omi Me

Complete Omi.me integration for memories, action items (tasks), and conversations. Full CRUD + sync capabilities for OpenClaw.

This skill appears to be what it claims: a simple CLI wrapper for Omi.me that stores and uses your OMI_API_TOKEN. Before installing or running the setup script: 1) Inspect the scripts yourself (they are plain shell) and confirm you trust the omi.me endpoints. 2) Note the bug in omi-cli.sh: it sets OMI_API_URL but uses API_URL when making requests; API calls will fail until corrected (change ${API_URL} to ${OMI_API_URL}). 3) The setup script can create symlinks in /usr/local/bin — avoid running it with sudo unless you intend to make the commands globally available; without sudo the symlink step will be skipped. 4) The token is stored at ~/.config/omi-me/token (or read from OMI_API_TOKEN); treat that file like any other secret (chmod 600 is already used by the script). 5) Because the skill makes network calls to https://api.omi.me/v1/dev, verify the domain and your API key's permissions before granting access. If you want to be extra cautious, run commands manually (bash ./scripts/omi-token.sh set/test and bash ./scripts/omi-cli.sh ...) rather than creating global symlinks.

Type: OpenClaw Skill Name: omi-me Version: 1.0.0 The OpenClaw AgentSkills skill bundle for Omi.me provides a legitimate CLI client for interacting with the Omi.me API. It handles API tokens securely by prompting for silent input and setting `chmod 600` permissions on the token file (`~/.config/omi-me/token`). The `omi-cli.sh` script performs standard CRUD operations against `https://api.omi.me/v1/dev` using `curl`, and the `setup.sh` script performs benign installation tasks like creating config directories, making scripts executable, and optionally creating symlinks in `/usr/local/bin`. There is no evidence of data exfiltration to unauthorized endpoints, malicious execution, persistence mechanisms, or prompt injection against the agent to perform actions outside the stated purpose.