The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to execute standard environment management and diagnostic commands such as pip freeze, conda env export, uv lock, renv::snapshot(), and Rscript. These operations are necessary to capture the current state of the computational environment.
[EXTERNAL_DOWNLOADS]: During the verification phase (Phase 3), the skill may trigger network activity to resolve or dry-run package installations using official package managers. This behavior is consistent with verifying that a lockfile is functional and targets well-known, public package registries.
[INDIRECT_PROMPT_INJECTION]: The skill represents an attack surface for indirect prompt injection because it ingests and processes user-controlled source code and configuration files to extract package names and random seeds.
Ingestion points: Reads files matching scripts/R/*.R, scripts/stata/*.do, scripts/python/*.py, and standard project configuration files like pyproject.toml or requirements.txt.
Boundary markers: Not explicitly mentioned in the instructions for data read from user files.
Capability inventory: The skill has access to Bash (for tool execution) and Write (for generating lockfiles and reports).
Sanitization: No specific sanitization or escaping of content extracted from user scripts is described before it is processed or written to outputs.

capture-environment