The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes standard CLI tools to gather data about skill history and status. This includes git log to check iteration history, gh pr list to find related pull requests, and shell commands like mkdir and redirection to manage output files.
[INDIRECT_PROMPT_INJECTION]: The skill has a surface for indirect injection as it reads and processes untrusted data from other skill files (SKILL.md) and logs to generate the article content.
Ingestion points: Reads content from skills/<name>/SKILL.md, memory/logs/, and memory/MEMORY.md.
Boundary markers: None explicitly defined in the instructions to separate external data from system instructions.
Capability inventory: File system write access (output/), shell command execution (git, gh, ./notify), and MCP tool invocation (higgsfield).
Sanitization: No specific sanitization or filtering is performed on the ingested content before it is processed by the model.
[SAFE]: The skill's operations are consistent with its stated purpose of technical writing and documentation. The external dependencies (git, GitHub CLI, and the Higgsfield MCP) are standard development tools and services.

skill-article