Bounded Loops and Graphs
github.com/qualixar/bounded-loops- Category
- AI Agents
- Rank
- No. 2098Tools index
Previous survey · No. 2003 ·
- Pricing
- Open Source
- Type
- TOOL
- Use case
- Agent Building
- GitHub
- 4 stars
- Latest release
- v0.7.6
- Date
About
Bounded Loops and Graphs is an open-source framework addressing the problem of AI agents claiming a task is "done" without that being verified. It ships 69 agent loops (65 requiring no API key) and 7 reference graphs, each checked by independent gates against declared bounds and producing verifiable receipts, and is accompanied by a research paper (arXiv:2609.27871).
At a glance
Bounded Loops and Graphs runs agent work against an external completion check and stops at a configured limit. It is worth studying if you want a repeatable retry loop with a record of how it ended.
Good fit
Developers who can define a pass condition the worker cannot modify.
Consider something else when
Design a better evaluation first if the only signal is the agent saying it finished.
Start here
Read the shipped keyless stub example. Check how a failing test differs from a passing run and what the record shows when the attempt limit is reached.
Cost and requirements
Apache-2.0 software with Python and npm distribution. The repository includes examples that require no model key.
Limits to know
The gate’s quality and isolation determine what a pass means. A verifiable run record does not establish that the task was fully specified.
Sources checked 2026-11-05. Source-based assessment.
Why it made the leaderboard
Bounded Loops & Graphs gives agent workflows declared bounds and verifiable receipts so a 'done' claim from an agent can actually be checked against independent gates.
Tags
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.