loop-design-check
Design a goal-oriented agent loop or review one for failure modes: spinning
by affaan-mFree
Description
loop-design-check is an agent skill from the ECC collection by affaan-m, listed on skills.sh. It helps design a goal-directed agent loop, or review an existing one, for the failure modes that make loops go wrong: spinning without progress, gaming the verifier against its goal, and running a wrong answer to completion. Its premise is that a language model has no built-in way to steer toward a goal across turns, so a feedback loop has to be wrapped around it, and the skill checks that the loop's goal is right and that it will not run away. It applies when a repeating task, such as write then test, or test then fix, is handed to an agent, or when an existing loop is suspected of cheating or spinning. It does not apply to a one-off task, a plain timer or poll, or to the mechanics of pipelines and recovery, which the listing assigns to autonomous-loops and continuous-agent-loop. A skill is a SKILL.md file of instructions that an AI agent loads when a task calls for it. It is free and open source, and the skills.sh listing shows all three security audits passing.…
Highlights
- Designs goal-directed agent loops
- Catches spinning and verifier gaming
- Reviews loops for runaway wrong answers
- Separates goal checks from loop mechanics
Ratings & Reviews
No ratings yet
Be the first to rate loop-design-check.
People also visit
dev-team
Simulate a collaborative dev team session where multiple role-based personas
loop-design-check FAQ
What does the loop-design-check skill do?
It helps you design a feedback loop around an agent, or review one, for failure modes such as spinning, gaming the verifier and finishing a wrong answer.
When should I not use it?
For a one-off task, or for a simple timer or poll, which the skill says should use /loop with no design needed.
Does it cover loop mechanics?
No. The listing assigns pipeline and recovery mechanics to autonomous-loops and continuous-agent-loop.
How do I install it?
Run: npx skills add https://github.com/affaan-m/ecc --skill loop-design-check
Is it free?
Yes. It is free and open source.
Are there security findings?
The skills.sh listing shows Gen Agent Trust Hub, Socket and Snyk all passing.
What are the best loop-design-check alternatives?
Popular alternatives to loop-design-check include eval-harness, microsoft-foundry and agent-eval. See all loop-design-check alternatives