The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data from Sentry issue resources (logs, breadcrumbs) and external web content scraped via Firecrawl. A malicious payload in an error log or a scraped website could attempt to manipulate the agent's logic during the root cause analysis or fix phases.
Ingestion points: Sentry issue details and web scraping via Firecrawl.
Capability: The skill has the ability to modify codebase files, call Sentry APIs, and execute shell commands.
Sanitization: There are no explicit boundary markers or sanitization procedures defined for processing external content.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill instructions direct the agent to search for and read sensitive files such as .env, .sentryclirc, and sentry.properties to find authentication tokens. While necessary for the skill's primary purpose of Sentry integration, accessing these files exposes all environment secrets to the agent.
[UNVERIFIABLE_DEPENDENCIES_AND_REMOTE_CODE_EXECUTION]: The documentation suggests installing a complementary tool using npx mushi-mushi. Executing packages via npx involves downloading and running code from a remote registry, which constitutes remote code execution. Although the package is associated with the vendor 'kensaurus', it represents an external and unverified dependency execution.

debug-sentry-monitor