canary-watch
Pass
Audited by Gen Agent Trust Hub on Apr 3, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [NO_CODE]: The skill consists exclusively of documentation and instructional content in markdown format. There are no scripts (Python, JavaScript, Shell, etc.) or binary files included in the skill package.- [SAFE]: No malicious patterns such as credential harvesting, remote code downloads, or privilege escalation attempts were identified. The skill's operations are limited to informational monitoring tasks.- [PROMPT_INJECTION]: The skill instructs the agent to analyze external web content, which presents a surface for indirect prompt injection (Category 8). An attacker who controls a monitored URL could potentially embed malicious instructions to influence the agent's behavior. However, this is an inherent risk of web-browsing capabilities and the skill does not contain instructions to bypass existing safety guardrails.
- Ingestion points: External URLs provided via the command arguments (SKILL.md).
- Boundary markers: None explicitly defined in the prompt instructions to isolate external content.
- Capability inventory: The skill requires the ability to browse the web (read console logs, performance metrics, and HTML content) and write to a local log file at
~/.claude/canary-watch.log. - Sanitization: No specific sanitization or filtering instructions for the retrieved website content are provided.
Audit Metadata