The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses subprocess.run to call external utilities pandoc (in scripts/contract_analyzer.py) and mmdc (Mermaid CLI, in scripts/mermaid_renderer.py). These invocations are necessary for the skill's primary functions: extracting text from Word documents and rendering business flowcharts to images. The calls use specific arguments targeting local file paths.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-provided contract files, which creates a potential surface for indirect prompt injection where an attacker could embed malicious instructions in the contract text.
Ingestion points: Contract text is read in scripts/contract_analyzer.py via pandoc and the document structure is unpacked in scripts/ooxml/unpack.py.
Boundary markers: The reference prompts in references/ do not currently define explicit boundary markers or "ignore" instructions for the ingested text.
Capability inventory: The skill has the ability to execute shell commands (pandoc, mmdc) and write files to the local system (reports, annotated documents, and images).
Sanitization: The skill follows security best practices by using defusedxml in scripts/document.py and scripts/utilities.py to prevent XML External Entity (XXE) attacks. It also normalizes Mermaid syntax in scripts/mermaid_renderer.py before execution.

contract-review