The Agent Skills Directory

[SAFE]: The skill's functionality is centered on local content optimization using Python scripts provided in the package. It adheres to a strictly local execution model.
[COMMAND_EXECUTION]: The skill utilizes three Python scripts (format_picker.py, post_linter.py, and repurpose_splitter.py) to automate content analysis. These scripts only use standard library modules (argparse, json, re, hashlib, os) and perform deterministic logic such as regex-based linting and text splitting.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data, such as articles or talk transcripts, to generate social media posts.
Ingestion points: scripts/post_linter.py and scripts/repurpose_splitter.py accept external text via CLI arguments or file inputs.
Boundary markers: There are no specific delimiters or markers used to isolate ingested content from instructions within the scripts' logic.
Capability inventory: The scripts are limited to local text analysis, console output (JSON/human-readable), and local file management for a content ledger (.linkedin-ledger.json). They do not possess network capabilities or the ability to execute arbitrary code.
Sanitization: Input data is processed as raw strings for regex matching and hashing; however, the skill's instructions mandate that a human author must review every line before publication, providing a critical security checkpoint.

linkedin-content