mwcHorus
v1.0.1Build, maintain, and extend Horus (local-first tech/event intelligence terminal). Use when working on Horus relay ingestion, RSS/source pipelines, macro data...
⭐ 0· 210·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Suspicious
medium confidencePurpose & Capability
The skill's name/description match the instructions: it is a local-first Horus terminal skill and correctly instructs how to run relay/UI and wire the OpenClaw gateway. However, the SKILL.md expects runtime env vars and local repo layout (.env, OPENCLAW_SESSION_KEY, J7 credentials, data files under ~/workspace/horus/horus-relay/data/) while the registry metadata declares no required env/config paths. That mismatch should be resolved or explained.
Instruction Scope
The instructions explicitly direct the agent to read many local files (~/workspace/horus/horus-relay/data/*), auto-check data on prompts, and append durable facts to ~/workspace/horus/MEMORY.md. The subagent rules mandate automatic reads ('do not wait for extra prompting') which gives the agent broad, automatic access to local workspace data — this is expected for a local-intel tool but is privacy-sensitive and should be explicit in the skill's metadata.
Install Mechanism
No install spec and no code files are included in the skill package — lowest install risk. The runtime commands in SKILL.md (npm install, npm run dev) operate on the user's workspace code, so the risk is running code from the repository, not installing arbitrary external artifacts via the skill.
Credentials
SKILL.md references several environment variables (OPENCLAW_SESSION_KEY, HOST, PORT, OPENCLAW_BASE_URL/OPENCLAW_TOKEN, J7_USERNAME/J7_PASSWORD, VITE_RELAY_URL) and relies on .env files, yet the skill declares no required env variables or config paths. Requiring gateway/session keys and upstream credentials is plausible for this purpose, but the omission in metadata is an incoherence and opaque to an installer — it could lead to accidental exposure if not handled carefully.
Persistence & Privilege
The skill will read and also append to MEMORY.md in the workspace (the SubagentSKILL.md's 'Memory requirement (mandatory)' instructs this). It does not request always:true and does not appear to modify other skills. Writing to the user's workspace memory file is expected for durable memory but should be made explicit to users before enabling the skill.
What to consider before installing
This skill is largely coherent with a local Horus dev/runtime workflow, but it asks the agent to read and automatically act on local workspace data and to rely on environment secrets that the metadata does not declare. Before installing or enabling autonomous use: 1) Review the horus repository in ~/workspace/horus yourself (especially .env.example, any .env files, and code under horus-relay) to ensure no sensitive secrets are present; 2) Confirm which environment variables the agent will need (OPENCLAW_SESSION_KEY, OPENCLAW_TOKEN, J7 credentials, HOST/PORT, VITE_RELAY_URL) and store them securely (do not commit .env); 3) If you plan to allow the agent to run npm install / npm run dev, run those commands manually in a sandbox first to verify the code; 4) If you do not want automatic data reads/writes, do not allow autonomous invocation or ask the author to remove the 'auto-check' mandate; 5) Ask the skill author to update registry metadata to list required env vars and config paths (and to justify automatic reads/ MEMORY.md writes) — if they do, the skill could be upgraded to benign. If you are unsure, run this skill only in an isolated environment or decline until metadata is clarified.Like a lobster shell, security has layers — review code before you run it.
latestvk976wbdywwrrz9t4nj16gfwk9982dyw8
License
MIT-0
Free to use, modify, and redistribute. No attribution required.