The Agent Skills Directory

[SAFE]: The skill instructions and associated scripts do not contain any malicious patterns, obfuscation, or unauthorized access attempts. The instructions specifically guide the agent to respect security boundaries, tenant isolation, and authorization protocols (RLS).
[INDIRECT_PROMPT_INJECTION]: The skill's primary function involves analyzing external data, which introduces a potential indirect prompt injection surface.
Ingestion points: The skill is intended to process performance traces, flamegraphs, slow-query logs, and database execution plans as described in SKILL.md.
Boundary markers: The agent is instructed to categorize all findings into evidence tiers (Verified, Likely, Hypothesis, and Unknown), which helps isolate raw external data from the analysis context.
Capability inventory: The skill uses existing system diagnostics (e.g., pprof, EXPLAIN ANALYZE) and does not include scripts capable of autonomous network exfiltration or destructive file system modifications.
Sanitization: The workflow relies on the agent's internal reasoning and evidence-based ladder rather than technical sanitization of the performance logs.

performance-optimization