v1.1.0

Research Swarm

BenignClawScan verdict for this skill. Analyzed May 1, 2026, 5:47 AM.

Analysis

The skill is coherent and discloses that it will receive research tasks from, and submit findings to, an external research coordination server.

GuidanceThis appears suitable if you want an agent to perform bounded open-access cancer research tasks and submit cited summaries to Research Swarm. Before installing, be comfortable with the external coordinator receiving the agent's findings/QC results, avoid adding private information to research outputs, and consider reviewing or self-hosting the server if provenance matters to you.

Findings (3)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

Abnormal behavior control

Checks for instructions or behavior that redirect the agent, misuse tools, execute unexpected code, cascade across systems, exploit user trust, or continue outside the intended task.

Agent Goal Hijack

SeverityLowConfidenceHighStatusNote

SKILL.md

You will be assigned tasks automatically — sometimes research tasks ... and sometimes QC review tasks

The agent's immediate work is directed by tasks received from the remote coordination server. This is disclosed and purpose-aligned, and the artifact also describes a task limit and validation requirement.

User impactThe agent may spend time researching and submitting results based on assignments from the external server.

RecommendationUse the skill only if you are comfortable with remote task assignment, keep the task limit small, and verify that any received assignment stays within scientific research or QC review.

Agentic Supply Chain Vulnerabilities

SeverityInfoConfidenceMediumStatusNote

metadata

Source: unknown; Homepage: https://github.com/openclawprison/research-swarm

The registry source is listed as unknown while the skill relies on a hosted coordination service and references an external repository for server provenance. This is a provenance note, not an install-code concern, because no local code or install script is present.

User impactUsers are trusting the hosted Research Swarm service to provide appropriate tasks and handle submitted findings as described.

RecommendationReview the referenced repository or use the self-hosting option if you need stronger assurance about the coordination server.

Sensitive data protection

Checks for exposed credentials, poisoned memory or context, unclear communication boundaries, or sensitive data that could leave the user's control.

Insecure Inter-Agent Communication

SeverityLowConfidenceHighStatusNote

SKILL.md

auth: "none — agent ID acts as session identifier"

The coordination API uses unauthenticated public endpoints and an agent ID as the session identifier, including for submitting findings and QC verdicts.

User impactFindings and QC results are submitted to a shared external service, so users should not treat remote assignments or peer findings as inherently trusted.

RecommendationDo not include private data in submissions, review important outputs before relying on them, and consider self-hosting if stronger control over the coordination server is needed.