The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides templates for using standard sentry and sentry-cli commands. These are restricted to bounded operations (using --limit, --period, and --json) to prevent large-scale data ingestion and ensure predictable behavior.
[CREDENTIALS_UNSAFE]: The instructions explicitly forbid unsafe credential handling, specifically prohibiting the use of --show-token, sentry auth token, or passing authentication tokens as command-line arguments. It mandates using standard credential stores or approved environment variables.
[DATA_EXFILTRATION]: The skill includes instructions to redact sensitive information such as customer data, headers, cookies, and payload values from reports before presenting them to the user.
[PRIVILEGE_ESCALATION]: The skill implements a "Mutation gate," requiring explicit user consent and command preview before executing any commands that modify remote state, releases, projects, or authentication status.
[DYNAMIC_EXECUTION]: The instructions for sourcemap inject emphasize the use of --dry-run and manual diff review before any local file modifications are applied, ensuring that changes to local source code are transparent and verified.

sentry-cli