The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests raw text from LinkedIn exports (CSV/XLSX) and local markdown drafts to perform automated classification, resonance scoring, and strategy calibration. This content is untrusted and could contain instructions meant to influence the agent's logic.
Ingestion points: The script scripts/build_archive.py reads post text from local *.md files and LinkedIn Shares.csv exports.
Boundary markers: No explicit boundary markers or "ignore instructions" delimiters are used when processing the ingested text.
Capability inventory: The agent has the capability to write to the local file system (posts.jsonl, ANGLES.md), send messages via Telegram, and deploy dashboards to Cloudflare Pages.
Sanitization: No sanitization or escaping of the external content is observed before it is used for classification or reporting.
[COMMAND_EXECUTION]: The skill requires the agent to execute a provided Python script (build_archive.py) to process and merge data. It also suggests using shell commands like find to determine the script's local path.
[EXTERNAL_DOWNLOADS]: The skill instructions specify the installation of the openpyxl Python library from the official package registry to enable Excel file parsing.

content-track