The Agent Skills Directory

[COMMAND_EXECUTION]: The skill requires the agent to execute local Python scripts, specifically scripts/quality_check.py and scripts/sop_checklist.py, to perform automated validation and ensure adherence to the standardized reporting process.
[EXTERNAL_DOWNLOADS]: The report template in references/html_skeleton_template.html includes a script tag that fetches the Chart.js library from cdn.jsdelivr.net, which is a well-known and established public content delivery network.
[DYNAMIC_EXECUTION]: The main SKILL.md file contains Python code snippets that the agent is instructed to execute locally to verify report integrity, including chapter anchor presence and ID uniqueness.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to process and summarize data from untrusted external sources, which creates a potential surface for indirect prompt injection.
Ingestion points: Data is retrieved from the web and patent databases through Web Search MCP and Patsnap MCP tools as specified in the Phase 2 workflow.
Boundary markers: The instructions do not define specific delimiters or security-focused framing for the interpolation of external content into the agent's reasoning process.
Capability inventory: The skill has the capability to execute local scripts and write to the file system (generating the HTML reports).
Sanitization: There are no documented requirements for sanitizing or escaping the data fetched from external sources before it is written into the HTML report or processed by the agent.

tech-insight-report