The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes a bash script (scripts/demo-checkpoints.sh) used for managing git-based checkpoints. This script uses standard commands like git, sed, sort, and head to tag and navigate project states. This is a legitimate functional component of the skill's purpose and does not exhibit malicious patterns.
[DATA_EXPOSURE]: The skill provides extensive advice on avoiding data exposure, specifically regarding credentials in .env files, browser history, terminal scrollback, and personal notifications. It recommends using synthetic data instead of anonymized real data to prevent re-identification. The instructions actively promote secure practices for managing secrets and sensitive information during presentations.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided demo details (claims, constraints, tech stack) to generate a structured markdown runbook. While this constitutes a data ingestion surface, the context is the generation of a planning document for the user's own use, and the instructions focus on technical engineering rather than executing untrusted commands. Capability inventory includes local file management and markdown generation.

developer-live-demo-design