The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses subprocess.run to execute a local Node.js script (scripts/zeitgeist-sample.js) for firehose data collection and a Python interpreter to run keyword extraction routines. These operations are core to the skill's functionality.
[DYNAMIC_EXECUTION]: In scripts/bsky.py, the extract_keywords function generates a Python script string that is executed via subprocess.run. The stopwords parameter is interpolated into this string without sanitization, creating a potential path for code injection if an attacker-controlled string is passed to this parameter.
[INDIRECT_PROMPT_INJECTION]: The skill retrieves untrusted content from the Bluesky network through search and feed functions. This content is then used for keyword extraction and image transcription, posing a risk of indirect prompt injection.
Ingestion points: External data is ingested via functions in scripts/bsky.py that call Bluesky's searchPosts, getAuthorFeed, and getFeed endpoints.
Boundary markers: There are no explicit delimiters or protective instructions used to isolate untrusted post content when it is passed into downstream processing tools.
Capability inventory: The skill possesses capabilities for network interaction and local command execution.
Sanitization: Post content is processed as raw text without filtering or escaping before being used in keyword extraction scripts.

browsing-bluesky