The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill directs the agent to use uvx disambiguate --lint. This command utilizes uvx to fetch and execute the disambiguate package directly from the Python Package Index (PyPI) at runtime without a pinned version or integrity check.
[COMMAND_EXECUTION]: The agent is instructed to invoke the uvx command-line utility within the project environment to perform linting tasks on glossary markdown files.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of external data which could contain malicious instructions.
Ingestion points: The skill reads user-supplied terms, edge-case scenarios, and existing project source code to inform the domain model.
Boundary markers: There are no explicit instructions or delimiters provided to the agent to distinguish between data and instructions when processing these inputs.
Capability inventory: The agent has the capability to read and write files to the docs/glossary/ and docs/adr/ directories, allowing potentially poisoned content to be persisted in the repository.
Sanitization: No sanitization or validation logic is specified for the content extracted from the code or user input before it is written into markdown files.

domain-modeling