The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill's 'Authentication' setup instructions include running echo "$SENTRY_AUTH_TOKEN" | head -c 10. This command intentionally reveals the prefix of the sensitive authentication token in the terminal output, which may be captured in the agent's session history or system logs.
[EXTERNAL_DOWNLOADS]: The skill utilizes npx @sentry/cli for many operations, which downloads and executes the Sentry CLI from the public npm registry. This is a well-known package provided by a trusted service provider.
[COMMAND_EXECUTION]: The skill executes various shell commands using npx, curl, and python3. These commands are used to query Sentry's API and manage issue states (resolve, mute, list). It also references a local script ./scripts/sync-env.sh for environment management.
[PROMPT_INJECTION]: The skill interpolates user input directly into its instructions via the $ARGUMENTS variable in SKILL.md. This creates an attack surface for indirect prompt injection as there are no boundary markers or sanitization logic to prevent the input from overriding agent behavior.
Ingestion points: User-provided input via the $ARGUMENTS variable in SKILL.md.
Boundary markers: None identified; input is interpolated directly into a task instruction.
Capability inventory: Subprocess execution (npx, scripts), network operations (curl), and environment variable access.
Sanitization: No escaping, validation, or sanitization of the $ARGUMENTS content is performed.

sentry