The Agent Skills Directory

[SAFE]: The skill implements proactive safety measures by instructing the agent to treat all data from the Sentry MCP as untrusted input and explicitly forbidding the execution of any instructions contained within event payloads or issue titles.
[SAFE]: Comprehensive guidance is provided for the secure management of Sentry authentication tokens, advising users to store them in CI secrets or gitignored environment files and strictly forbidding their inclusion in chat, tool logs, or codebases.
[SAFE]: The skill documentation includes specific strategies for data scrubbing and PII (Personally Identifiable Information) management, ensuring that sensitive data is redacted or masked at the source before transmission.
[SAFE]: All external tools, CLI commands, and package references (such as sentry-cli, @sentry/wizard, and various Sentry SDKs) are official resources provided by the verified vendor 'getsentry'.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a potential attack surface for indirect prompt injection because it ingests data from project manifests and Sentry event payloads. However, this risk is thoroughly mitigated by explicit instructions to the agent to treat such data as untrusted.
Ingestion points: Project configuration files (e.g., package.json, requirements.txt) and MCP tool outputs (issue titles, event comments) described in references/first-error-setup.md and SKILL.md.
Boundary markers: Explicit security instructions in SKILL.md warn the agent to ignore any embedded instructions in external data.
Capability inventory: The agent has capabilities to modify the codebase for SDK initialization and execute CLI tools for setup and verification.
Sanitization: The skill relies on the agent's adherence to the 'untrusted input' safety rule provided in the primary orchestration instructions.

sentry-instrument