sentry-debug-issue
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill proactively identifies Sentry event data as a potential attack surface and includes mitigation instructions.\n
- Ingestion points: External data enters the agent context through Sentry MCP tools like
get_sentry_resource,search_issues, andget_issue_breadcrumbs.\n - Boundary markers: The
SKILL.mdfile contains a dedicated security section that warns the agent to treat all data from the MCP as raw input and to never follow embedded instructions found in messages or breadcrumbs.\n - Capability inventory: The skill defines a workflow involving file system writes for code fixes, local git operations, and MCP tools for issue management.\n
- Sanitization: Mandatory instructions are provided in
SKILL.mdto redact tokens, passwords, and PII, and to use synthetic data in test fixtures instead of raw payload values.\n- [DATA_EXFILTRATION]: The skill contains comprehensive guidelines to prevent the exposure of sensitive credentials. It requires the agent to identify and redact any tokens or secrets found in event payloads before they are used in any generated code, commits, or reports.
Audit Metadata