The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill uses npx claude-flow@alpha in multiple lifecycle hooks (pre, post, pre-search, etc.) to execute framework-specific tasks, which involves downloading and running packages from the NPM registry.
[REMOTE_CODE_EXECUTION]: Through the use of npx, the agent downloads and executes the claude-flow package logic at runtime to perform its core operations.
[COMMAND_EXECUTION]: The skill executes shell commands defined in the hooks and Workflow sections of SKILL.md. These commands use variable interpolation (e.g., ${description}, ${results}, ${summary}) that could be vulnerable to command injection if the input data for these variables (such as the code being analyzed) contains malicious shell characters.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its primary function of analyzing untrusted external codebases.
Ingestion points: The agent ingests and processes source code files during the "Initial Scan" and "Deep Analysis" phases in SKILL.md.
Boundary markers: No specific delimiters or "ignore instructions" markers are used to separate the analyzed code from the agent's internal instructions.
Capability inventory: The agent has the capability to execute shell commands (via npx), store and retrieve persistent data in memory, and send notifications.
Sanitization: There is no evidence of sanitization or filtering applied to the code content before it is processed or used in analysis results.

agent-code-analyzer