The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill attempts to download and install the cc-conversation-search package from the Python Package Index (PyPI) or via uv if the tool is not already present on the system. This package is an external dependency not included in the standard agent toolkit.
[INDIRECT_PROMPT_INJECTION]: The skill processes human messages from past conversations to perform its analysis, which creates a surface for indirect prompt injection if those logs contain malicious commands.
Ingestion points: Human-role messages extracted from conversation JSONL files (SKILL.md).
Boundary markers: The skill does not implement delimiters or explicit instructions to isolate the ingested data from the analysis agent's instructions.
Capability inventory: Analysis is conducted using Task agents, and the primary skill has access to powerful tools including Bash, Write, and Read.
Sanitization: The skill filters for whitespace and automated tool results but does not sanitize or escape the content of the human messages before they are processed by agents or interpolated into the HTML report.
[COMMAND_EXECUTION]: The skill utilizes the Bash tool to perform environment checks, install software, initialize search databases, and automatically trigger the open command to display the generated HTML report.

ai-fluency