The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill retrieves data from external APIs and references remote libraries in its output.
It communicates with https://redfox.hk/story/api/x/hotAccount/rankList to fetch entrepreneur ranking data.
It communicates with https://redfox.hk/story/api/x/userTimeline to retrieve tweet timelines for analysis.
The generated HTML reports reference the html2canvas library from cdn.jsdelivr.net to enable image export functionality.
[COMMAND_EXECUTION]: The skill uses subprocesses for local workflow management and report viewing.
In scripts/fetch_rank.py, subprocess.run is used to invoke the generate_report.py script once data is fetched.
In scripts/generate_report.py, subprocess.run (using open or xdg-open) or os.startfile is used to launch the user's default web browser to view the generated HTML report. These operations target internally generated file paths.
[INDIRECT_PROMPT_INJECTION]: The skill serves as a surface for indirect prompt injection as it ingests and summarizes content from X (Twitter).
Ingestion Points: Data enters via the redfox.hk API, including fields like fullName, biography, and tweet text (processed in scripts/fetch_tweets.py and scripts/fetch_rank.py).
Boundary Markers: The SKILL.md includes explicit instructions for the agent to treat the analysis as "based on public tweets" and "only for reference," but does not use strict delimiters for all interpolated data.
Capability Inventory: The skill has network access (via urllib.request) and script execution capabilities (via subprocess.run).
Sanitization: The skill uses html.escape (via the esc helper in scripts/generate_report.py) to sanitize external content before inserting it into the HTML report, which effectively mitigates cross-site scripting (XSS) risks in the resulting document.

x-entrepreneur-rank