The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses local shell commands including git, find, grep, and wc to perform reconnaissance on the repository. These commands are restricted to read-only analysis of the file system and version history to identify complexity and churn hotspots.
[PROMPT_INJECTION]: The skill identifies and processes source code from the repository, which presents an attack surface for indirect prompt injection if malicious instructions are embedded in the codebase.
Ingestion points: Sub-scanners spawned in Phase 3 read arbitrary source files within the user-selected scope.
Boundary markers: The instructions provided to the sub-scanners do not include specific delimiters or warnings to ignore instructions found within the code comments or content being analyzed.
Capability inventory: The skill possesses the ability to create project issues using the bd tool and spawn additional sub-agents for analysis.
Sanitization: No explicit sanitization of code content is performed before it is analyzed by the language model, though the requirement for user triage before filing issues serves as a manual sanitization step.

refactor-finder