loop-maker
EricTechPro/loop-maker
Portable agent skill that interviews you, then scaffolds a self-running loop with verifier and human gate
loop-maker is a portable agent skill for Claude Code, Codex, Hermes and OpenClaw that asks seven questions and then scaffolds a self-running autonomous loop with a separate verifier, a state file and a human approval gate.
What loop-maker does
The wizard asks seven questions one at a time: goal, trigger, discovery, action, verification, state and human gate. The goal must be a checkable true or false condition, verification must come from a separate judge, state must live outside the chat, and irreversible actions must ask a human first. Question seven cannot be skipped. A progress bar, breadcrumb and blueprint box are rendered live by scripts/loop_progress.py, a dependency-free Python 3 helper.
The output is a ready-to-run loop folder under loops/ containing SKILL.md, STATE.md, a verifier script and a stop condition. It targets three common mistakes: an agent that judges its own work, changing state leaked into the skill file, and a loop with no stop condition. The skill scaffolds loops but is not a runtime, so the result runs as a separate skill or cron job. You start it with /loop-maker or by describing an automate, schedule or monitor task.
Key features
- Seven-question wizard covering goal, trigger, discovery, action, verification, state and human gate
- Scaffolds SKILL.md, STATE.md, a verifier script and a stop condition
- Non-skippable human gate for irreversible actions
- Live progress bar and blueprint box from a dependency-free Python helper
- Starts with /loop-maker or from a plain task description
When to use it
- Setting up a scheduled monitoring agent with its own verifier and budget limit
- Building a content pipeline or data-sync loop that stops at a defined condition
- Adding an approval step before an automated outreach sequence sends anything
Who it is for: People building autonomous agent loops who want the verifier, state and stop condition designed up front.
How it fits with Hermes Agent
Described as a portable skill that works in Hermes alongside Claude Code, Codex and OpenClaw.
How to install loop-maker
These commands are copied from the project's README. Check the repository for the latest steps before you run them.
git clone https://github.com/EricTechPro/loop-maker ~/.claude/skills/loop-maker
LOOP_MAKER_SKILLS_DIR=~/my-project/.claude/skills ./install.shRequirements: Python 3 for the bundled scripts/loop_progress.py helper
FAQ
What is loop-maker?
loop-maker is a portable agent skill that interviews you with seven questions and scaffolds an autonomous loop. The result includes a verifier, a state file, a stop condition and a human gate.
Does loop-maker work with Hermes Agent?
Yes, the README says it works in Claude Code, Codex, Hermes and OpenClaw. It generates loops but does not run them, so the loop executes as a separate skill or cron job.
How do I install loop-maker?
Clone the repository into your skills folder, for example ~/.claude/skills/loop-maker, or run ./install.sh with LOOP_MAKER_SKILLS_DIR pointing at a custom skills directory. Then start it with /loop-maker.
Similar skills for Hermes Agent
All skillsHermes Agent skills for live HTML artifact preview, nine-lens deep research and benchmark testing
Neeeophytee Agent StylebooksPortable Agent Skills that give AI agents editorial systems drawn from public style guides
konglong87 super-pmSkill pack of 56 product-management workflow entries for Claude Code, Cursor, Codex and OpenCode
SupercmoHQ SuperCMO SkillsOpen-source agent skills that produce UGC videos, ad videos and product photography from a product photo
LearnPrompt Carl SkillsCollection and registry of real-world AI workflow skills for creators, reusable in Hermes and other agents
AtlasOmnia Hermes Agent Custom PackIndependent pack of skills, plugins, integrations and scripts for Hermes Agent by AtlasOmnia
Related guides: What is Hermes Agent? · SOUL.md for Hermes Agent: what it is and how to write one