Back to skill
Skillv1.0.0
ClawScan security
Team · ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
BenignMar 8, 2026, 7:09 PM
- Verdict
- benign
- Confidence
- medium
- Model
- gpt-5-mini
- Summary
- The skill is a high-level, instruction-only specification for orchestrating multi-agent teams and is internally consistent with its stated purpose, but it is vague and could give an agent broad discretion when executed.
- Guidance
- This skill is a conceptual framework and contains no code or external requests, so it appears coherent with its stated purpose. Before installing or enabling it, consider: (1) limit the agent's permissions (network, other skills, file system) so vague orchestration rules can't trigger broad actions; (2) do not provide financial or privileged credentials to any agent invoking orchestration without explicit need; (3) test behavior in a sandboxed environment to observe what the agent does when asked to 'recruit' or 'distribute rewards'; (4) require explicit human approval for any actions that involve external APIs, payments, or modifying other skills. If you need stronger assurance, ask the skill author for a concrete operational spec showing what APIs or other skills the orchestration will call and what safeguards are in place.
Review Dimensions
- Purpose & Capability
- okName/description (autonomous orchestration framework) align with the SKILL.md content: a conceptual engine (role allocation, conflict resolution, reward routing). The lack of required binaries, env vars, or installs is consistent with an instruction-only design.
- Instruction Scope
- noteThe SKILL.md contains only high-level architecture primitives and a lifecycle; it does not instruct the agent to read files, call external endpoints, or access secrets. However the instructions are intentionally vague (open-ended orchestration language) and would leave implementation choices to the agent, which may lead to broad or unexpected actions at runtime if the agent is permitted to act autonomously.
- Install Mechanism
- okNo install spec and no code files — lowest-risk delivery model (instruction-only). Nothing is downloaded or written to disk by the skill itself.
- Credentials
- okThe skill does not request any environment variables, credentials, or config paths. That is proportionate for a purely conceptual/orchestration spec, though real orchestration implementations would typically require additional service credentials which this skill does not request.
- Persistence & Privilege
- noteFlags show default behavior (user-invocable, agent may invoke autonomously). Autonomous invocation is normal, but combined with the skill's vague orchestration remit it could enable broad multi-agent behavior depending on the agent's other permissions — the skill itself does not request permanent/always-on placement or modify other skills.