fix-sentry-issues

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external error logs and stack traces without establishing explicit boundary markers or isolation, creating a potential surface for indirect prompt injection if an attacker can manipulate error payloads.
  • Ingestion points: Sentry issue details and events retrieved via get_issue_details and search_issue_events (SKILL.md).
  • Boundary markers: Absent. No specific delimiters or validation guidelines are provided to prevent the agent from executing instructions embedded in log strings.
  • Capability inventory: Executes git commands (git checkout, git pull), performs Axiom CLI queries, updates Sentry issue states via mcp__sentry__update_issue, and makes arbitrary network fetches.
  • Sanitization: Absent. No verification or filtering is applied to the retrieved Sentry attributes or variables.
  • [COMMAND_EXECUTION]: The skill instructs the agent to run shell commands locally for version control operations and tool queries.
  • Evidence: Executes git checkout main && git pull && git checkout -b fix/<descriptive-name> and axiom query "['shiori-events'] | where traceId == '<traceId>'" -f json (SKILL.md).
  • [EXTERNAL_DOWNLOADS]: Instructs the agent to perform fetch() requests on external URLs extracted from Sentry logs during the reproduction phase.
  • Evidence: Instructs the agent to "fetch() the URLs that timed out" to verify understanding (SKILL.md), which poses a minor risk of Server-Side Request Forgery (SSRF) if the URL in the log is attacker-controlled.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 09:16 AM
Security Audit — agent-trust-hub — fix-sentry-issues