The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructions define a workflow for creating and configuring agent 'specialists' and 'skills' based on user requests. This process allows external input to be persisted into the agent's execution environment, which could be used to override behavior if malicious instructions are provided.
Ingestion points: User customization requests processed by the agent as described in SKILL.md.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are specified for the configuration files.
Capability inventory: The skill enables modification of agent prompts in .feynman/agents/ and functional scripts in the skills/ directory.
Sanitization: There are no instructions for validating or sanitizing the user-provided content before it is applied to the workbench configuration.
[NO_CODE]: The skill consists entirely of instructional text and does not contain embedded scripts, binaries, or references to external code dependencies.

customize