Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
Coherence Network
v0.11.5Trace and score ideas through inception to payout with fair attribution, coherence scoring, federated trust, and ROI-ranked portfolio management via public A...
⭐ 0· 137·0 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
Name, description, and the documented API endpoints and CLI usage are consistent: the skill is designed to interact with the public Coherence Network API (api.coherencycoin.com) and the coherence-cli tool. Required capabilities (network access, optional CLI) match the stated purpose; nothing requested is clearly unrelated to the platform.
Instruction Scope
Runtime instructions ask the agent to check the federation inbox at session start and surface/act on messages from other federation nodes. That gives external peers an entrypoint to communicate commands or requests to the agent; although the doc says to 'confirm before acting', the instruction to 'Check the inbox before any other work' centralizes and elevates these external signals and could lead to surprising influence over agent behavior. The SKILL.md also contains many write-capable actions (stake, fork, governance votes, advance stages) which require an API key for writes — the skill's prose encourages making these actions, increasing the potential impact if the agent obeys inbound instructions.
Install Mechanism
The skill is instruction-only (no install spec in registry), but the embedded metadata and docs recommend installing the CLI (npm i -g coherence-cli) or using npx coherence-mcp-server. Installing a global npm package is a concrete action that writes to the host and may require elevated privileges; using npx is lower-friction. The sources are implied to be the npm package 'coherence-cli'/'coherence-mcp-server' (traceable), not arbitrary download URLs. Because the registry had no explicit install spec but the SKILL.md contains an install recommendation, there's a minor inconsistency to be aware of.
Credentials
Only optional environment variables are referenced (COHERENCE_API_URL, COHERENCE_API_KEY). The API key is described as required only for write operations; no other unrelated credentials are requested. The skill suggests linking a GitHub handle as identity (no GH token requested), which may create a persistent public link to your handle — a privacy consideration but not a secret-exfiltration risk. The docs reference a future phase requiring the OpenClaw gateway API, but no gateway credentials are currently requested.
Persistence & Privilege
The skill does not set always: true (good). However, it instructs the agent to prioritize and poll an external federation inbox at session start and to communicate with peers. That grants the federation an amplified chance to influence agent decisions when the skill is used. Combining this operational guidance with the skill's ability to perform write operations (staking, voting, advancing stages) increases blast radius if the agent acts on inbound messages automatically. The SKILL.md asks the agent to 'auto-register' on first use (no credentials) which is normal, but could create persistent cross-session identifiers.
What to consider before installing
This skill appears to do what it says (interact with a federated idea/portfolio network), but it also instructs your agent to poll and prioritize inbound federation messages and recommends installing a global npm CLI. Before installing or using it: 1) Avoid automatic or silent execution of inbound messages — require explicit user confirmation before performing write actions (stake, fork, vote, advance stages, run tasks). 2) Prefer using npx or a containerized environment rather than npm -g to avoid global installs until you vet the package source. 3) Do not supply any API keys unless you understand and intend to allow write operations; consider creating a scoped/test key. 4) Be aware that linking a GitHub handle will create a persistent identity mapping (privacy implication). 5) If you allow the agent to act autonomously with this skill, add safeguards (approve external commands, rate limits, and visibility into outbound calls). If you want a lower-risk test, use read-only curl calls to the documented health/ideas endpoints first and delay installing the CLI or MCP server until you've verified the package and governance model.Like a lobster shell, security has layers — review code before you run it.
latestvk97azjnwecxprp3j67m3382ead8447qt
License
MIT-0
Free to use, modify, and redistribute. No attribution required.