The Agent Skills Directory

[PROMPT_INJECTION]: The skill explicitly instructs the agent to prioritize instructions found in an external file over its own definitions.
Evidence: "read its AGENTS.md before anything else. It holds the studio's commands and rules, and wins where it differs from this skill."
[EXTERNAL_DOWNLOADS]: The skill clones a software studio repository from the author's GitHub account.
Evidence: Fetches resources from https://github.com/yahyabedirhan/explainer-studio to set up the environment.
[COMMAND_EXECUTION]: The skill triggers shell commands to build, render, and preview video projects.
Evidence: Executes npm run dev and follows setup/render instructions provided in the cloned repository's README.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to instructions embedded in the external repository it processes.
Ingestion points: The AGENTS.md file within the cloned explainer-studio repository.
Boundary markers: None; the skill explicitly grants the external file authority to override its own instructions.
Capability inventory: Git cloning, file system scaffolding, script writing, voice generation, and shell command execution (npm run dev).
Sanitization: No sanitization or validation of the instructions read from AGENTS.md is specified.

make-explainer