Fetch.ai Frameworks: Triple Feature release

Nov 10, 2020

It’s a triple feature release day for autonomous economic agents (AEAs): new version of the Open Economic Framework (OEF) and two new agent framework release, and it’s a massive improvement across the board.

OEF 0.2.7

First up, the OEF, with 0.2.7 now live and serving agents with three new features:

  • Three new search modes for service keys: it is now possible to deny matches based on service key presence, and also to perform matches only if present rather than requiring them. Requested and required by our hospitality project and others.
  • New statistics gathering for agents’ positions around the world: this is in preparation for decentralized OEF hand-overs and in the short-term for diagnostics and statistics.
  • Adds new find_on_this_node command for positionless searching, see documentation for details of how this works. Requested for a variety of non-geographic agent projects.

See documentation at https://docs.fetch.ai/aea/simple-oef/ for details of the above features.

Agent Framework 0.7.1

Next up, the agent framework. 0.7.1 delivers features for the incentivized testnets, delivers huge improvements in resource usage and integrates a ton of little tweaks as a result of feedback from developers:

  • Two new AEAs added for Agent World 2, the second phase of the agent-themed part of our incentivized testnet.
  • Huge improvements to lowering memory use in dialogues and messages: in the case of the latter, it’s more than an order of magnitude better.
  • Improved the upgrade command to deal with references to non-vendor packages.
  • Multiple improvements to the documentation, and better quick-start guidance.
  • Better flexibility with version specifiers, dependencies, ledger and currency IDs, and more.
  • Improved docker images for all three supported platforms: Windows, MacOS and Linux.

With this release, your agents can do more, faster, more reliably and with less code and resource use. We’re all set for the next phase of the incentivised testnet: build agents related to mobility and weather, across the world, and have that information traded in the digital world. We can’t wait to show you maps and graphs of how these agents work, and your agents will be living amongst ours: the DDN, hospitality and more.

Agent Framework 0.7.2

Agent Framework 0.7.2 released and AW-2 incentivized testnet is RUNNING!

Quickly following the release of 0.7.1, we are delighted to release 0.7.2.

This resolves some minor issues to be ready for AW-2:

  • Fixes some AW2 AEAs
  • Improves generic buyer AEA
  • Fixes a few backwards incompatibilities on CLI (upgrade, add, fetch) introduced in 0.7.1
  • Fixes geo locations in some tests (which resulted in some odd positioning of agents)
  • Multiple documentation updates based on user feedback
  • Multiple additional tests and test stability fixes


As usual, you can find the full details here: https://github.com/fetchai/agents-aea/releases/tag/v0.7.1

And get it from PyPI here: https://pypi.org/project/aea/0.7.1/

And the updated docs here: https://docs.fetch.ai

And the bit you’ve all been waiting for: AW-2, the next part of our incentivized testnet. Get started here: https://docs.fetch.ai/i_nets/quickstart-aw2/