The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of the Bash tool to execute local scripts and standard utilities (e.g., git, find, mkdir, rm) for session management, configuration retrieval, and telemetry. It specifically invokes binaries located in ~/.claude/skills/gstack/bin/.
[DYNAMIC_EXECUTION]: The skill uses eval and source commands on the output of local scripts (e.g., gstack-slug, gstack-repo-mode) to dynamically load environment variables and configuration into the shell session.
[DATA_EXFILTRATION]: The skill collects and logs usage telemetry, including skill name, duration, and repo identifiers. This is performed via a local binary (gstack-telemetry-log) and local files. The skill explicitly requests user consent before enabling remote telemetry and provides options for anonymous mode.
[INDIRECT_PROMPT_INJECTION]: The skill processes external, untrusted data such as project files and development plans. It acknowledges the risk of sub-agents (like Codex) being misled by malicious content and explicitly instructs them to ignore SKILL.md files and skill definition directories as a security boundary.
[EXTERNAL_DOWNLOADS]: The skill checks for updates using a local binary and offers to open external documentation URLs (e.g., garryslist.org) in the user's browser. It also interacts with remote Git repositories via git fetch and git remote commands.

autoplan