The Agent Skills Directory

[OBFUSCATION]: The skill implements a 'Data Fusion' feature that embeds the source YAML data as a base64-encoded blob within the generated HTML report. This is a declared architectural decision in builder/fuse.py designed to ensure reports are self-contained and reproducible. While it utilizes encoding, it serves a legitimate technical purpose rather than concealing malicious instructions.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data from web research to populate the newsletter content.
Ingestion points: Research results from current web developments (defined in SKILL.md).
Boundary markers: None explicitly defined for the interpolation of search results into the agent context.
Capability inventory: File system writes in builder/build_report.py and browser automation via Playwright in builder/capture_screenshots.py.
Sanitization: The Jinja2 rendering engine in builder/build_report.py is configured with autoescape=True to mitigate potential cross-site scripting (XSS) in the generated reports.
[COMMAND_EXECUTION]: The utility script builder/capture_screenshots.py uses the Python subprocess module to execute node for automated screenshot generation using Playwright. This capability is intended for developer asset regeneration and testing rather than standard user-facing newsletter generation.
[EXTERNAL_DOWNLOADS]: The skill relies on standard, well-known third-party dependencies including PyYAML and Jinja2 for Python, and playwright for Node.js-based screenshot automation. These are used according to best practices via pyproject.toml and requirements.txt.

newsletter-ai