The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8). It ingests data from external configuration files which could contain malicious instructions designed to manipulate the subagents used for auditing.
Ingestion points: The skill reads .lsp.json files and plugin.json manifests from the project directory.
Boundary markers: No explicit boundary markers or instructions to ignore embedded commands are used when passing file content to the subagents.
Capability inventory: The skill utilizes Bash, Read, Glob, Grep, and Task tools.
Sanitization: No evidence of sanitization or validation of the input file content is present before it is processed by the agent.
[COMMAND_EXECUTION]: The skill uses the Bash tool to perform its auditing tasks. This capability represents a potential risk if an attacker successfully uses indirect prompt injection to execute arbitrary shell commands via the auditor subagents.

audit-lsp