The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to read React component source files to extract prop types and component names. While source code is technically untrusted data that could contain malicious instructions in comments (e.g., trying to redirect the file-writing process), the skill's narrow focus on structural analysis for test generation significantly limits the impact of such an attack.
Ingestion points: Component source files found in static/app/components/ (SKILL.md).
Boundary markers: Absent.
Capability inventory: Reading local files and writing new .snapshots.tsx files to the filesystem.
Sanitization: Absent.
[EXTERNAL_DOWNLOADS]: The skill mentions imports from @sentry/scraps. As the author of the skill is 'getsentry', this is recognized as a legitimate vendor-owned resource used for internal project dependencies.
[COMMAND_EXECUTION]: The skill suggests using Glob or Grep to find files based on the $ARGUMENTS provided. While this involves command execution, it is standard behavior for developer-focused AI agents within a controlled project environment and does not pose a risk of unauthorized privilege escalation or data exfiltration in this context.

generate-snapshot-tests