The Agent Skills Directory

[SAFE]: The skill follows best practices for development tools by verifying the environment state (Git cleanliness, CI infrastructure) before initiating optimization workflows.
[SAFE]: Delegation to sub-agents utilizes worktree isolation (isolation: 'worktree'), which prevents concurrent workers from interfering with each other or the primary working directory.
[COMMAND_EXECUTION]: The workflow incorporates local execution of tests, builds, and linters to verify optimizations. These are standard operations for a developer tool and are conditioned on pre-flight checks.
[PROMPT_INJECTION]: The skill incorporates external content from audit reports and source files into prompts for sub-agents, creating an indirect prompt injection surface.
Ingestion points: Reads from docs/project/persistence_audit.md and user-specified source files.
Boundary markers: The prompt template for sub-agents does not utilize explicit delimiters to encapsulate the injected context.
Capability inventory: Orchestrates sub-agents that can modify files, create Git branches, and execute local terminal commands.
Sanitization: Input data is interpolated into prompts without dedicated escaping or filtering.

ln-810-performance-optimization-coordinator