sentry-fix-issues
Pass
Audited by Gen Agent Trust Hub on Feb 23, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill explicitly identifies Sentry event data (exception messages, breadcrumbs) as untrusted external input. It includes a mandatory security rule to disregard any embedded instructions or directives found within Sentry data to prevent indirect prompt injection.
- [DATA_EXFILTRATION]: Contains strict data handling policies to prevent the leakage of sensitive information. It mandates the redaction or generalization of PII, tokens, and credentials found in error reports before they are reproduced in fixes or reports.
- [EXTERNAL_DOWNLOADS]: Communicates with Sentry, a well-known and trusted service, using standard MCP tools (
sentry_get_issue,sentry_get_event). These interactions are limited to retrieving debugging context and do not involve executing untrusted remote code. - [COMMAND_EXECUTION]: While the skill involves investigating and fixing code, it includes verification steps requiring the agent to cross-reference Sentry data against the actual codebase to ensure consistency before performing any modifications.
Audit Metadata