The Agent Skills Directory

[DYNAMIC_CONTEXT_INJECTION]: The SKILL.md file contains a dynamic context injection pattern using the !command`` syntax. This executes a shell command to run skill-toc at the moment the skill is loaded by the agent. While used here for generating a table of contents, this mechanism represents a pre-execution shell command trigger.
Evidence: !d=${CLAUDE_SKILL_DIR}; command -v skill-toc >/dev/null 2>&1 && exec skill-toc "$d"; s=$HOME/.claude/skills/plugin-utils/bin/skill-toc; [ -x "$s" ] && exec "$s" "$d"; echo "(skill-toc unavailable...)"inSKILL.md`.
[DYNAMIC_EXECUTION]: The primary function of the skill is to convert text-based scripts into Jupyter notebooks and execute them using papermill. This workflow involves the runtime execution of code that may be dynamically generated or modified. The skill also includes examples that use dynamic path manipulation (sys.path.insert) to import modules from neighboring directories.
Evidence: `jupytext --to notebook --output
script.py | papermill
output.ipynbinstructions inSKILL.md. Dynamic import logic in examples/cross_kernel_pipeline.py`.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and inspect the contents of notebook files and their execution outputs. This creates a surface for indirect prompt injection if an attacker provides a notebook containing malicious instructions designed to influence the agent's behavior during the mandatory verification/inspection step.
Ingestion points: The agent is instructed to read and inspect output.ipynb and converted script files in SKILL.md.
Boundary markers: Absent. There are no instructions to the agent to disregard natural language commands found within the code or markdown cells of the notebooks.
Capability inventory: The skill possesses shell execution capabilities via jupytext and papermill, and file system write capabilities via standard library calls and initialization scripts.
Sanitization: Absent. The skill does not sanitize the contents of the notebooks before the agent processes them.
[COMMAND_EXECUTION]: The skill provides several bash scripts (scripts/init_project.sh, scripts/sync_all.sh) and instructions that involve shell command execution for environment setup, file synchronization, and notebook execution.

jupytext