The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Downloads the Bun runtime installer from the well-known service domain at https://bun.sh/install during the setup procedure.
[REMOTE_CODE_EXECUTION]: Executes the downloaded Bun installation script using Bash after verifying its integrity with a hardcoded SHA-256 checksum (bab8acfb046aac8c72407bdcce903957665d655d7acaa3e11c7c4616beae68dd). This follows security best practices for remote script execution.
[COMMAND_EXECUTION]: Frequently invokes local GStack binaries from ~/.claude/skills/gstack/bin/ to manage configuration, check for updates, and track session state. It also employs dynamic execution patterns, such as eval and source, on the output of these local utilities to manage project environment variables.
[DATA_EXFILTRATION]: Contains features for logging telemetry (usage statistics) and syncing project artifacts (design docs) to remote infrastructure. These features are strictly opt-in; the skill explicitly requests user permission via the AskUserQuestion tool before any data is transmitted externally.
[PROMPT_INJECTION]: Analyses project-specific files such as CLAUDE.md, TODOS.md, and previous design documents to gather context for brainstorming sessions. This ingestion of untrusted local data represents a surface for indirect prompt injection, although the skill includes instructions to sub-agents to ignore potentially malicious content within these files.

office-hours