The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions to download the Sentry CLI installation script from https://cli.sentry.dev/install, which is an official domain for the well-known Sentry service.
Evidence: Found in SKILL.md and references/COMMAND_PATTERNS.md.
[REMOTE_CODE_EXECUTION]: The skill facilitates the execution of a remote shell script via a pipe to bash for tool installation.
Evidence: curl https://cli.sentry.dev/install -fsS | bash in SKILL.md.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted external data from Sentry environments (such as issue titles, error messages, and log content) which could potentially contain malicious instructions targeting the agent.
Ingestion Points: Data enters the context through commands like sentry issue list, sentry log list, sentry event view, and sentry trace view (multiple reference files).
Boundary Markers: The instructions do not define clear delimiters or specific warnings to the agent to disregard instructions embedded within the retrieved Sentry data.
Capability Inventory: The agent has the ability to perform network operations, file system configuration modifications (via .sentryclirc), and destructive actions like sentry project delete or sentry issue archive (as documented in references/project.md and references/issue.md).
Sanitization: There is no evidence of content filtering or sanitization of the Sentry API responses before they are processed by the agent.

sentry-cli