The Agent Skills Directory

[COMMAND_EXECUTION] (MEDIUM): Executes multiple shell scripts from ~/.claude/skills/sentry-observability/scripts/, such as init_sentry.sh and create_alert.sh. These scripts are external to the skill bundle, meaning their specific logic and safety cannot be verified.
[EXTERNAL_DOWNLOADS] (LOW): Uses pnpm and npx to install and run packages from the npm registry (@sentry/nextjs, pino, posthog-js). This introduces a dependency on the supply chain security of these third-party modules.
[DATA_EXFILTRATION] (SAFE): Network operations using curl are performed against localhost for verification of the health endpoint, which does not pose an exfiltration risk.
[CREDENTIALS_UNSAFE] (SAFE): While the skill instructs the setup of sensitive environment variables like SENTRY_AUTH_TOKEN, it uses placeholders rather than hardcoding actual secrets.

fix-observability