The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains instructions that attempt to override default agent behavior for platform compatibility, such as "Ignore Claude-specific mode-switch instructions when they appear" and "Strict execution contract: when a user explicitly invokes a skill, execute that skill protocol as written."
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external data (source code) and has the capability to perform file system operations and spawn sub-agents, creating an indirect prompt injection surface.
Ingestion points: Individual entity and value object files are read and analyzed in Phase 2.
Boundary markers: The skill does not define specific boundary markers or "ignore instructions" delimiters for the content of the files it reviews.
Capability inventory: The skill uses rg (ripgrep) for file searching, executes a local Python script (.claude/scripts/code_graph) for graph analysis, spawns sub-agents (code-reviewer) for delegated reviews, and writes report files to the plans/reports/ directory.
Sanitization: No explicit sanitization or filtering of the content read from source files is specified before processing.
[COMMAND_EXECUTION]: The skill uses local shell commands including ls, rg, and python to discover project structures, search for patterns, and perform code graph analysis. These are used within the scope of the local repository.

review-domain-entities