BigHugger
sk Skill · affaan-m

loop-design-check

Design a goal-oriented agent loop, and review it for the ways loops go wrong — spinning and burning tokens, Goodhart-gaming the verifier, or running a wrong answer to completion. Two actions: (1) WRITE a loop — gate whether to build it, define a machine-decidable goal, pick the loop type, pick a skeleton; (2) REVIEW a loop — run it past five failure modes plus decidability, boundaries, fallback, judge independence,…

installs 8w
1,664
30-day movement
starts with the next reading
Related entries
2
Connections
0
JavaScript

A skill that guides an agent through designing a goal-oriented agent loop and reviewing an existing one. It covers a five-step write path (build gate, machine-decidable goal, loop type, skeleton, damping) and a review checklist of five failure modes plus red lines like judge independence and keeping judgment with the human.

Reach for it when you want an autonomous loop that won't spin, game its verifier, or run a wrong answer to completion.

Use it to

  • Design a new plan/build/judge loop from scratch
  • Check an existing loop against the five failure modes
  • Define a machine-decidable exit condition
  • Add damping against runaway or oscillating loops
  • Decide whether a task warrants a loop at all

For Developers building or auditing autonomous agent loops

Host repository
affaan-m/ECC
Installs, lifetime
1,800
Installs, 8 weeks
1,664
Host stars
261k
Host language
JavaScript
topicsagent-loopsloop-designverificationrunaway-preventionskillreview-checklist