Back to skill

Security audit

Rook CLI by TestMu AI

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rates this skill suspicious because its standard workflow may sync unspecified project data to a cloud service without disclosing the payload or requiring approval; the scan is mostly complete but has one unresolved reference.

Before installing, confirm where cloud synchronization sends data and what it includes, and avoid uploading sensitive project material unless you have reviewed and authorized the disclosure. The scan is partial because a referenced `agent.yaml` file was not found, so the evidence does not cover that reference.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

The workflow instructs cloud synchronization without disclosing the data sent to the third-party service.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

The standard run workflow directs the assistant to perform cloud synchronization with --yes, without identifying the uploaded data or requiring approval of that disclosure. Project testing material may contain sensitive information, and authorization to test an agent or spend credits does not necessarily authorize uploading that material to a third party; the actual payload is not established by this source.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · references/mcp.md (reported line 23)May include surrounding context.

md
a server `explore` finds already declared in the target. `approve` trusts a
`project` or `discovered` server after rook has read what it runs; those are
the only two origins that wait for it. A `user` or `local` server is used as
declared, with no approval step.

When `explore` finds `.mcp.json` in the target, those servers appear with
origin `discovered` and need `approve` before a scenario can exercise them.

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · references/profiles.md (reported line 12)May include surrounding context.

md
| Kind    | Give rook                      | Example                                                                                                    |
| ------- | ------------------------------ | ---------------------------------------------------------------------------------------------------------- |
| HTTP    | a curl                         | `curl http://127.0.0.1:9110/v1/triage -H 'content-type: application/json' -d '{"input":"look at T-1043"}'` |
| command | a command line with `{{goal}}` | `claude -p "{{goal}}"`                                                                                     |
| MCP     | a tool on a declared server    | see `mcp.md`                                                                                               |

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · references/profiles.md (reported line 19)May include surrounding context.

Authoring headlessly

bash
printf '%s\n' "curl http://127.0.0.1:9110/v1/triage -H 'content-type: application/json' -d '{\"input\":\"{{goal}}\"}'" > /tmp/profile.txt
rook profile add triage-http --from /tmp/profile.txt --yes
rook profile add refund-cli --command 'claude -p "{{goal}}"' --yes
rook profile test --goal "Say hello and nothing else."