Hesham A.

You can reconcile AI coding agent configurations with official docs, drive work from goal clarification through verified execution, stage and commit single atomic changes with approval, audit commits before pushes or pull requests, and run value-based cross-model reviews of plans, diffs, or decisions.

They target individual operators of AI coding harnesses such as Claude Code, Codex, OpenCode, and Hermes Agent who want disciplined, approval-gated change management. The author self-describes as a nerd rather than a developer, so the skills emphasize verification and human oversight.

The develop skill carries work from goal clarification through planning, execution, and verification while preserving resumable state. Verified repository changes are handed to commit for a single approved atomic commit, and publish reviews the exact commits a push, release, or pull request would expose.

Spar provides value-based cross-model review of a plan, diff, or decision through a read-only reviewer bridge. It lets a second model critique proposed changes without granting write access, adding an independent evaluation layer before commits or releases proceed.

Run eyrsync after relevant harness changes or client releases, or as a periodic review. It reconciles Claude Code, Codex, OpenCode, and Hermes Agent configuration, skills, permissions, and controls against official documentation, releases, local references, and runtime evidence.