The Agent Skills Directory

[SAFE]: The skill is a documentation-only resource that provides principles and checklists for performance engineering tasks such as latency measurement and database optimization.
[EXTERNAL_DOWNLOADS]: The skill references an evaluation artifact and a source repository on GitHub (github.com/jacob-balslev/skill-graph). These are static, informational links to the author's own repository and do not trigger automated code execution.
[INDIRECT_PROMPT_INJECTION]: The skill instructions involve analyzing external performance data (e.g., query plans, bundle sizes). While this represents an ingestion surface for potentially untrusted data, the skill's restricted tool access (Read, Grep) and lack of write or network capabilities prevent exploitation via data injection. Findings for this category:
- Ingestion points: Baseline measurements, query plans, and bundle analysis results are ingested into the agent context (SKILL.md).
- Boundary markers: None specified.
- Capability inventory: The skill is limited to the Read and Grep tools (SKILL.md).
- Sanitization: No explicit sanitization or validation of the ingested measurement data is described.

performance-engineering