Native Sentry

Read Sentry issues, events, and production errors via the Sentry REST API. Use when the user wants to inspect errors, list recent issues, get stack traces, o...

This skill appears to do exactly what it says: read-only access to Sentry via the REST API. Before installing, confirm you provide a Sentry token with only read scopes (project:read, event:read, org:read). Be cautious if you override the base URL (SENTRY_BASE_URL or --base-url) because the token will be sent to that host — only point it at trusted Sentry instances. Avoid using --no-redact in shared or logged environments because it can expose PII and stack traces. If you need a higher assurance, review the included scripts/sentry_api.py yourself (it is small and uses only stdlib networking).

Type: OpenClaw Skill Name: native-sentry Version: 1.0.1 The skill bundle is classified as suspicious due to a potential shell injection vulnerability. While the `sentry_api.py` script itself employs good security practices like PII redaction, `argparse`, and `urlencode` to prevent injection within its own logic, the `SKILL.md` defines commands that take arguments (e.g., `ISSUE_ID`, `EVENT_ID`, `--query`) which, if directly interpolated from unsanitized user input by the OpenClaw agent into the bash command string, could lead to arbitrary shell command execution. This represents a significant vulnerability in the agent's interaction model with the skill, rather than intentional malice within the skill's code itself.