The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill implements a 'Project Baseline Bootstrap' process that ingests untrusted codebase data. It uses tools like 'rg' to list files and reads repository documentation (e.g., README, CONTRIBUTING) to infer project facts. This establishes an attack surface where malicious instructions embedded in a project's documentation could attempt to influence the agent's behavior.
[DYNAMIC_EXECUTION]: The skill instructions require the agent to execute a workspace helper script ('aegis-workspace-helper'). The path to this script is resolved dynamically at runtime from environment variables (AEGIS_WORKSPACE_HELPER) or local user configuration. This is used for legitimate workspace lifecycle management (e.g., directory structure checks).
[COMMAND_EXECUTION]: The skill relies on shell command execution for environment discovery and project indexing. Specifically, it uses git commands (e.g., 'git rev-parse', 'git status') to identify repository state and 'rg' (ripgrep) to discover project files. These operations are restricted to read-only discovery tasks during the initial phase.

using-aegis