The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes multiple shell commands, including thurview, git show, and npx. These commands are used to perform architectural analysis, generate code graphs, and manage document revisions.
[REMOTE_CODE_EXECUTION]: The skill employs npx -y thurview to download and execute the thurview package directly from the NPM registry. It also provides instructions for a thurview update command which fetches code from a remote source at runtime.
[DATA_EXFILTRATION]: Through the thurview publish command, the skill uploads local design metadata, anchored code snippets (peeks), and software map structures to a remote service, which generates a browser-accessible URL.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process instructions and guidance from external local files, creating a potential vector for indirect influence over agent behavior.
Ingestion points: The agent is instructed to read guidance from ~/.thurview/THURVIEW.md, the repository-level THURVIEW.md, and local authored files such as review.md and data.yaml.
Boundary markers: The instructions do not specify the use of delimiters or ignore-previous-instructions warnings for the content of these files.
Capability inventory: The skill possesses capabilities for shell execution (thurview, git) and network communication (thurview publish).
Sanitization: There is no evidence of sanitization or filtering applied to the content retrieved from these guidance files before processing.

thurview-design