The Agent Skills Directory

[SAFE]: The skill is designed to help users write and optimize Instagram captions using a local script called caption.py. No malicious patterns were detected.
[COMMAND_EXECUTION]: The SKILL.md file contains instructions for the agent to execute python3 caption.py. This is the intended primary purpose of the skill and is performed on local files using a script included in the skill package.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided caption drafts through the caption.py linter. While this represents a data ingestion surface, the skill lacks dangerous capabilities such as network access, file-writing to arbitrary paths, or privilege escalation, which mitigates the risk of exploitation.
Ingestion points: Reads text from local files (e.g., caption.txt, draft.txt) or via standard input.
Boundary markers: Not explicitly defined in the script, but the linter performs structured analysis on the input.
Capability inventory: Limited to reading files and printing analysis results to the console via caption.py.
Sanitization: The script uses regular expressions for pattern matching rather than executing the input text as code.
[DATA_EXPOSURE]: No hardcoded credentials, sensitive file paths, or network exfiltration patterns were found. The script operates entirely on the input provided by the user.
[EXTERNAL_DOWNLOADS]: The skill does not download external scripts, packages, or configurations at runtime. It relies solely on Python's standard library (argparse, json, re, sys, textwrap).

ig-caption