The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied requests and project-specific preferences stored in decisions.md to generate new, optimized instructions for the agent.
Ingestion points: The skill reads project preferences from decisions.md and takes raw user input to perform rewrites.
Boundary markers: There are no explicit boundary markers or sanitization logic defined to prevent instructions within the processed data from influencing the agent's logic, though this is inherent to the skill's purpose as a prompt engineer.
Capability inventory: The skill is designed to rewrite prompts and then execute the resulting output in the same turn.
Sanitization: No sanitization of external content is specified before interpolation into the prompt rewrite loop.
[EXTERNAL_DOWNLOADS]: The documentation provides instructions for users to manually install the skill via GitHub, which is a standard distribution method for AI agent extensions.
The repository https://github.com/Jakeschincariol/promptmaster-skill is the primary installation source for the skill's logic and configuration files.
[SAFE]: The skill uses local files for configuration and auditing, following standard patterns for prompt engineering tools.
The waste-patterns.json file contains benign prompt improvement guidelines focused on token efficiency and clarity.
Instructions focus on reducing conversation churn and improving deliverable specificity without requesting elevated privileges or network access.

promptmaster