The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from project manifest files such as pyproject.toml and pixi.toml during its detection phase. 1. Ingestion points: project manifests (e.g., pyproject.toml, pixi.toml, poetry.lock, uv.lock, environment.yml, requirements.txt). 2. Boundary markers: Absent. No explicit delimiters or instructions to ignore embedded instructions are used when reading these files. 3. Capability inventory: Execution of shell scripts and package management commands (pixi, uv, poetry, hatch, conda, pip). 4. Sanitization: Absent. Manifest content is not sanitized before being used in logic or displayed.
[EXTERNAL_DOWNLOADS]: The skill's scripts and documentation reference remote installation URLs for well-known developer tools. Evidence: scripts/install_agent_feature_pixi.sh and scripts/install_agent_feature_uv.sh contain echo messages referencing https://pixi.sh/install.sh | sh and https://astral.sh/uv/install.sh | sh. Context: These are provided as guidance for users and are not directly executed by the agent.
[COMMAND_EXECUTION]: The skill executes local bash scripts to manage dependencies and verify environmental layout. Evidence: scripts/install_agent_feature_*.sh and scripts/verify_layout.sh. Context: These commands are essential for the skill's primary function of environmental configuration.

python-env-manager