The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes various system commands and binaries to perform its optimization tasks. This includes calling a local orchestrator binary (./orchestrator) and language-specific toolchains such as go, cargo, npx, ruff, and pytest. While these are functional requirements for the skill, they involve direct execution of code in the local environment.
[INDIRECT_PROMPT_INJECTION]: The skill's architecture is susceptible to instructions embedded within the projects it analyzes. Maliciously crafted project documentation or source code could influence the agent's task generation phase.
Ingestion points: Phase 1 (SKILL.md) instructs the agent to read project files including README, CLAUDE.md, and source code.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to treat project content as untrusted or to ignore embedded instructions.
Capability inventory: The skill possesses extensive capabilities including file system writes, git branch management, and the execution of shell commands and local binaries.
Sanitization: The skill does not implement any visible sanitization or validation of the content read from the target project before using it to design optimization tasks.

auto-optimize