Openclaw Spirits
v1.3.2Companion spirit system for OpenClaw agents. Every user gets a unique spirit creature — deterministically generated from their identity (24 species, 5 rariti...
⭐ 2· 345·2 current·2 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name/description, assets (sprites.json, companion.json), and scripts (generate.js, render.js, display.js, soul.js) align with a deterministic companion generator. No unrelated credentials, binaries, or platform-level access are requested.
Instruction Scope
Runtime instructions ask the agent to run local node scripts, call the agent's LLM to generate the 'soul', and save companion data under {baseDir}/assets/companion.json — all consistent with the stated purpose. The skill also instructs the agent to occasionally initiate brief passive messages; that is scope-appropriate for a 'companion' skill but could be considered unsolicited behavior by some users.
Install Mechanism
No install spec provided (instruction-only from registry). Code and assets are packaged with the skill; there are no external download/install steps or third-party package pulls in the manifest.
Credentials
Skill declares no required env vars or credentials, which matches the instructions. It does require a user identifier (Feishu open_id, Telegram ID, Discord ID, or username) to deterministically generate a spirit — this means PII-like identifiers will be used as seed input and stored in an in-skill companion.json; consider privacy implications of using platform IDs.
Persistence & Privilege
always:false and model invocation allowed (platform default). The skill's guidance asks for occasional passive appearances (agent-initiated one-liners), but this autonomy is not combined with broad credentials or system changes. No evidence the skill modifies other skills or system-wide configs.
Assessment
This package appears to be what it says: a deterministic 'spirit' companion that runs entirely from local scripts and the agent's LLM. Before installing, consider: (1) Privacy — the skill seeds generation with a user ID (open_id, Telegram/Discord ID, or username) and stores companion.json in the skill directory; if you don't want those identifiers used/stored, don't install. (2) Passive messages — the skill explicitly directs the agent to sometimes send short unsolicited one-liners; if you prefer no autonomous chatter, disable or avoid enabling the skill. (3) Review soul.js and other scripts yourself (or request their full source) to confirm they do not make external network calls or require API keys; the SKILL.md claims no network activity, but verifying the actual scripts eliminates residual doubt.Like a lobster shell, security has layers — review code before you run it.
companionvk97cmp67qn66vm9yf19e7sh9vn8450e7creaturevk97cmp67qn66vm9yf19e7sh9vn8450e7engagementvk97cmp67qn66vm9yf19e7sh9vn8450e7familiarvk97cmp67qn66vm9yf19e7sh9vn8450e7gamificationvk97cmp67qn66vm9yf19e7sh9vn8450e7identityvk97cmp67qn66vm9yf19e7sh9vn8450e7latestvk97cmp67qn66vm9yf19e7sh9vn8450e7openclawvk97cmp67qn66vm9yf19e7sh9vn8450e7petvk97cmp67qn66vm9yf19e7sh9vn8450e7spiritsvk97cmp67qn66vm9yf19e7sh9vn8450e7virtual-petvk97cmp67qn66vm9yf19e7sh9vn8450e7
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
