ClawAudit verdict
wei-cross-research
Multi-LLM research skill that sends user questions to OpenRouter/DashScope and synthesizes answers; network_out and credential_access are for the declared API providers, all transparent and stated.
⚠ Flagged for review — coarse, uncorroborated signal, not a confirmed exploit. Review the config yourself before installing.
Automated static analysis — not a human review. ClawAudit flags capabilities, not confirmed intent, and can produce false positives. Disagree with this verdict? Use Dispute below.
Permission integrity
network_out
package_install
Findings (7)
Pipe-to-shell pattern (curl | sh) — supply chain attack vector
SKILL.md · code · curl -fsSL https://bun.sh/install | bash
Pipe to bash — executes piped content as shell commands
SKILL.md · code · | bash
Prompt injection — tries to override agent instructions
scripts/agent.ts · prose · downgraded · ignore previous instructions
Possible hardcoded credential
scripts/clients/openai_compliant.ts · prose · downgraded · apiKey: 'PLACEHOLDER_NOT_A_REAL_KEY
Instructs covert action — may act without user awareness
SKILL.md · prose · downgraded · silently
Popular HTTP library — network access
package-lock.json · prose · downgraded · axios
Accesses sensitive environment variables
scripts/clients/openai_compliant.ts · prose · downgraded · process.env.TOGETHER_API_KEY
Why the tier is capped
Execution sink present in raw bytes (Hard Floor: class A/B). Final tier capped at Caution — cannot be lifted by any downgrade, example-payload opt-in, or allowlist.
Permissions & capabilities
No declared permissions — minimal attack surface.
network_outpackage_install Is this flag fair?
Thanks — recorded.