Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
Consilium
v1.1.0Your personal board of AI advisors — the only skill that uses truly different AI models (not one model role-playing). Get better answers to hard questions by...
⭐ 2· 502·0 current·0 all-time
bySergey Morozik@morozsm
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
The name/description match the behavior in SKILL.md: it orchestrates 3–5 different models and synthesizes responses. It does not request unrelated credentials or binaries. The claim that models come from different providers is a function of the user's model configuration (agents.defaults.models); the skill does not add extra secrets or external services. Note: example model names (e.g., github-copilot/gpt-5.2) assume the agent environment already has access to those provider models — the skill does not provide or request those credentials itself.
Instruction Scope
Runtime instructions are narrowly scoped: spawn parallel sub-agents, collect their session history, synthesize, and save a small council-panel.json to workspace root. The skill explicitly warns that user queries are sent to each configured model provider (expected behavior). There are no instructions to read unrelated system files, search user filesystem, or transmit data to unexpected endpoints. Small note: the skill relies on session auto-archiving per OpenClaw settings — users should confirm archive policy if they care about persistence.
Install Mechanism
Instruction-only skill with no install spec and no code files. Nothing is downloaded or written to disk by an installer beyond the explicit save of council-panel.json at runtime, so install risk is minimal.
Credentials
No environment variables, keys, or config paths are required by the skill itself. It relies on models already configured in the agent (which implies provider credentials exist elsewhere). This is proportional to its purpose; it does not request unrelated secrets or broad system access.
Persistence & Privilege
The skill saves council-panel.json to the workspace root (panel model names and slot assignments only) and uses sub-agent session memory / auto-archive for panel responses. This is reasonable for reuse but means panel configuration and possibly archived session content may persist according to agent settings — users should check workspace and archive retention policies. always:false and normal autonomous invocation settings are appropriate.
Assessment
This skill is an orchestrator that will send your question to each model provider you have configured. Before installing/using it: 1) confirm you trust the model providers in your agents.defaults (their credentials remain in your environment, not provided by the skill); 2) be aware each council run multiplies API calls (costs and rate limits); 3) check your OpenClaw session archival and workspace retention settings if you don't want panel responses or the council-panel.json to persist; and 4) if you expect access to specific providers, make sure those models are actually available in your agent configuration — the skill won't add credentials for you.Like a lobster shell, security has layers — review code before you run it.
latestvk971mbrnkbpas989csa0236xk181skj8
License
MIT-0
Free to use, modify, and redistribute. No attribution required.