Overview
Agent-FleetOps is a pattern library for orchestrating multi-agent fleets, built from scratch without frameworks like CrewAI or LangGraph. It addresses failure modes such as error cascades, argument loops, and unverified agent trust through drift guards, deterministic apply, and review panels. The repository includes a Textual terminal monitor (fleet-tui) for observing local and cloud model fleets, with a hermetic test suite and strict one-way pipeline. It also provides skills, templates, and tools for agent discipline, evaluation integrity, and model routing. The project emphasizes binding every check to a live target that can fail, as stated: "A report that cannot fail is indistinguishable from a report that passed." It includes 6093 tests in guard/tests/ to prove guards can fail, and a two-box throughput log with 67 measurements over 22 model tags. The design is validated in production via ParaKit.
Pricing
Free
Detailed plans have not been confirmed in our catalog. Check the official website for current limits and billing terms.
Visit WebsitePrices and limits may change. Confirm the currency, billing period, seat minimum and usage caps on the official website.
Use Cases
- Teams use Agent-FleetOps to monitor local and cloud model fleets in a single terminal interface.
- Developers use it to enforce agent discipline through generalized procedures and templates.
- Engineers use it to verify that every guard can fail via mutation testing.
- Operators use it to serialize writes across multiple agents with a driver-lock protocol.
- Researchers use it to conduct research as a chain of inspectable artifacts with independent verification.
- Teams use it to benchmark model throughput across hardware tiers with a two-box operating log.
Features
AI assistant
Natural language interface
Workflow automation
Reviews
No reviews yet. Be the first to share your experience!