← ClaudeAtlas

spec-herdr-reviewlisted

Run visible multi-agent spec reviews in Herdr through interactive agent CLIs such as Claude, Codex, Copilot/GLM, or Grok/Composer. Supports two modes — a spec review (the design, before implementation) and an implementation review (the implemented diff vs the spec, after the calling model has built it). Ask for reviewers first if none are specified. Requires HERDR_ENV=1.
stewartcelani/skills · ★ 2 · AI & Automation · score 70
Install: claude install-skill stewartcelani/skills
# Spec Herdr Review Use this skill to ask one or more external agent CLIs for a review from visible Herdr panes in a per-project reviews workspace, separate from the dev-server and working workspaces. This skill drives **interactive agent CLIs through Herdr's agent surface**, not headless prompt mode. Herdr starts the real CLI UI in a pane, `herdr agent prompt` submits the instruction, and Herdr's lifecycle states tell you when each reviewer is finished. Read the `herdr` skill first if you are not already fluent in the CLI. This skill covers the review workflow; `herdr` covers the command surface, ID rules, and safety rules. ## Preflight Herdr is required. Verify you are inside a Herdr-managed pane before doing anything else: ```bash test "${HERDR_ENV:-}" = 1 ``` If the check fails, say you are not running inside Herdr and stop. Do not fall back to another multiplexer. ## Two Review Modes This skill runs in one of two modes. **Decide the mode before doing anything else**, because it changes the filenames, what the request file points the reviewer at, and what the reviewer is allowed to read/run. ### Mode A — Spec review (BEFORE implementation) Review the **design itself**, verified against the **existing codebase**. The reviewer judges whether the plan is correct, complete, and safe to build — and a core part of that is **reading the current code to confirm the spec's claims are true**: that the files, line references, integration points, current behaviors, and ass