The Agent Skills Directory

[DYNAMIC_EXECUTION]: The provided Python script (scripts/tracking_commit.py) utilizes importlib.util.spec_from_file_location to dynamically load a local module (scripts/wordcount_core.py) at runtime based on a computed path. This pattern of dynamic loading is a technique that can be used to execute arbitrary local code if the file path resolution is compromised.
[COMMAND_EXECUTION]: The skill directs the agent to execute shell commands to invoke local scripts, specifically node scripts/check-outline-contract.js and python3 scripts/tracking_commit.py. These scripts are used to perform project validation and commit tracking data to the filesystem, representing a core functionality of the skill that involves automated execution of provided logic.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted novel text provided by users to generate project assets such as summaries, character snapshots, and story outlines. This creates a surface for indirect prompt injection, where malicious instructions hidden within the story text could be executed by the agent during its analysis and extraction tasks.
Ingestion points: User-supplied novel files (.txt, .md) or text pasted directly into the chat.
Boundary markers: The process relies on chapter headers to segment text but does not implement robust isolation or specific instructions to the model to ignore potential injection attempts in the source text.
Capability inventory: The skill has the capability to write files to the project directory and execute scripts via the agent's shell access.
Sanitization: There is no explicit evidence of sanitization or filtering logic applied to the novel content before it is processed by the agent's summarization and extraction capabilities.

story-import