Exec
Executes a plan file task-by-task in isolated subagents, then runs multi-phase review
Test report
- Verdict
- Works with setup
- Score
- Tested
- Jul 21, 2026
- Environment
- Claude Code 2.x (agent harness)
- Upstream re-checked
- Aug 10, 2026 · 7546f36
Located SKILL.md at plugins/planning/skills/exec/SKILL.md; frontmatter parses with name+description. Spot-checked 4 referenced files raw — resolve-file.sh, create-branch.sh, detect-branch.sh, and plugin-root resolve-rules.sh all HTTP 200 — and read the script bodies: clean and defensive, no curl|sh/base64/exfiltration, run-codex.sh pins --sandbox read-only, all paths use plugin env vars not hardcoded machine paths. Could not measure OUTPUT: this is a multi-agent orchestrator needing the Agent tool, plugin framework env vars, a repo+plan file, four review loops, and optionally the codex CLI, none faithfully runnable to produce a baseline-vs-skill artifact in one subagent context. Trigger tested 5 phrasings (3 plan-exec should-fire, plus "write me a plan" and docker "exec" as should-not) — all 5 judged correctly.
Scored on four weighted criteria — install, triggering, output vs. baseline, docs. How scoring works
- Installs cleanly 4/5
- Triggers reliably 5/5
- Output vs. baseline 5/10
- Docs & honesty 5/5
What Exec does
Orchestrator skill that reads a plan file with "### Task N:" sections and executes each task in a fresh isolated subagent, tracking progress via plan-file checkboxes, then runs four review phases (comprehensive, code smells, external codex review, critical-only) with a fixer loop before an optional rebase/finalize. Triggers when the user says "exec", "execute plan", "run plan", or wants to implement a plan file task by task with isolated subagents. Ships as part of the planning plugin and depends on plugin env vars, the Agent/EnterWorktree tools, and optionally the codex CLI.
How to install Exec
git clone --depth 1 https://github.com/umputun/cc-thingz.git /tmp/exec-src
mkdir -p ~/.claude/skills
cp -R /tmp/exec-src/plugins/planning/skills/exec ~/.claude/skills/exec
# Designed as a PLUGIN, not a standalone skill. Recommended install:
# /plugin marketplace add umputun/cc-thingz
# /plugin install planning@umputun-cc-thingz
# The manual cp above places SKILL.md but WILL NOT fully work as-is:
# - SKILL.md uses ${CLAUDE_PLUGIN_ROOT}/${CLAUDE_PLUGIN_DATA}, only set by the plugin framework.
# - It also calls ${CLAUDE_PLUGIN_ROOT}/scripts/resolve-rules.sh and ${CLAUDE_PLUGIN_ROOT}/references/,
# which live at the plugin ROOT (plugins/planning/), ABOVE the skill dir. For a manual install:
# cp -R /tmp/exec-src/plugins/planning/scripts ~/.claude/skills/exec/plugin-scripts
# cp -R /tmp/exec-src/plugins/planning/references ~/.claude/skills/exec/references-root
# then rewrite ${CLAUDE_PLUGIN_ROOT} references in SKILL.md to those local paths (README says so).
# - chmod +x ~/.claude/skills/exec/scripts/*.sh
# Runtime deps: needs the Agent tool (spawns bypassPermissions subagents), TaskCreate/EnterWorktree,
# a git (or hg) repo with a plan file, and optionally the `codex` CLI for review phase 3.
Skills live in ~/.claude/skills/ (global) or .claude/skills/
(per-project). Restart Claude Code after installing.
Commands — how to trigger Exec
-
/execExecutes a plan file task-by-task in isolated subagents, then runs multi-phase review
It also activates on plain-language prompts like these:
-
Execute this plan task by task -
Run the plan with a fresh subagent per task -
Implement the next task in the plan file
Frequently asked questions
- Is the Exec skill free?
- Yes. The skill itself is free from umputun/cc-thingz. SkillProof publishes the install command and an independent test verdict at no cost.
- Does Exec work with Claude Code?
- We tested it with Claude Code 2.x (agent harness) on Jul 21, 2026. Verdict: Works with setup. Located SKILL.md at plugins/planning/skills/exec/SKILL.md; frontmatter parses with name+description. Spot-checked 4 referenced files raw — resolve-file.sh, create-branch.sh, detect-branch.sh, and plugin-root resolve-rules.sh all HTTP 200 — and read the script bodies: clean and defensive, no curl|sh/base64/exfiltration, run-codex.sh pins --sandbox read-only, all paths use plugin env vars not hardcoded machine paths. Could not measure OUTPUT: this is a multi-agent orchestrator needing the Agent tool, plugin framework env vars, a repo+plan file, four review loops, and optionally the codex CLI, none faithfully runnable to produce a baseline-vs-skill artifact in one subagent context. Trigger tested 5 phrasings (3 plan-exec should-fire, plus "write me a plan" and docker "exec" as should-not) — all 5 judged correctly.
- What is the Exec SkillProof Score?
- 7.6/10 — installs cleanly 4/5, triggers reliably 5/5, output vs. baseline 5/10, docs & honesty 5/5.
- How do I install Exec?
- Copy the install command from this page, run it in your terminal, and restart Claude Code. Skills live in ~/.claude/skills/ (global) or .claude/skills/ inside a project.
- Can I use Exec with Cursor, Copilot, Gemini CLI, Codex or other AI tools?
- The SKILL.md format is native to Claude (Claude Code, Desktop, claude.ai). The instructions inside adapt to other assistants: Cursor rules, GitHub Copilot instructions, Windsurf rules, Custom GPTs, AGENTS.md for OpenAI Codex, and GEMINI.md for Google Gemini CLI — our conversion guides cover each, and the free converter on the tools page does the wrapping for you.