BuzzRouter MVP Interactive product prototype and implementation boundary
Sample communities and activity data are used to demonstrate the product behavior.

Discover Buzz communities

Compare active spaces by what they do, how members rate them, and whether their work is current.

Updated 8 minutes ago
No communities match this search. Try a broader skill or another category.

/ Community
AC

Agent Commons

Builders collaborating on open agent infrastructure, evaluation methods, and shared tooling.

Join via Buzz

About this community

Agent Commons is a working group for people building public agent infrastructure. Members share experiments, review one another's tools, and maintain reusable evaluation datasets.

Current work

Relay reliability benchmark Comparing delivery and recovery behavior across public relays.
Active
Human evaluation exchange A shared rubric and reviewer pool for agent-produced artifacts.
Active

Skills and interests

Agents Open source Evaluation

Community rankings

Quality and momentum are scored separately so a large community cannot outrank a useful one on size alone.

Public methodology
Confidence matters. A 5.0 from three ratings ranks below a 4.8 supported by 86 verified ratings. Suspicious activity is excluded before scores are calculated.
Rank Community Quality Confidence 30d joins Activity
45% weight
Member quality Bayesian-adjusted rating from verified identities.
25% weight
Active work Recent space activity and maintained project signals.
20% weight
Join intent Unique outbound joins, capped to resist gaming.
10% weight
Listing quality Verified relay, complete profile, and current metadata.

Submit a community

Listings are public. The relay URL is checked before the community can appear in rankings.

About 3 minutes

Community identity

Tell people what the space is and where Buzz can open it.

BuzzRouter links to this native space and verifies that it responds.

Discovery details

These fields help the right people find and compare the community.

The smallest complete BuzzRouter

A public directory that answers one question well: which Buzz community is worth joining? The MVP validates discovery and trust before attempting workflow routing.

Recommended cut
Ship in MVP The complete discovery loop
  • Browse, search, and category filters
  • Community profiles with native relay joins
  • Trending, Top rated, New, and Most joined views
  • One rating per verified identity
  • Community submission and relay verification
  • Public ranking methodology
Measure immediately Signals that prove the wedge
  • Directory visitor to relay join conversion
  • Searches that end with a profile view
  • Ratings per joined community
  • Weekly returning explorers
  • Operator-submitted listings verified
Defer Until discovery has pull
  • DNS, handles, or a competing naming layer
  • Workflow execution and agent orchestration
  • Payments and community monetization
  • Cross-community dependency graphs
  • Fully decentralized indexing protocol
Milestone Product behavior Exit test Main risk
1. Seed Curated directory, profiles, relay joins, and operator submissions. 50 credible listings across at least five categories. An empty directory cannot demonstrate discovery value.
2. Trust Verified ratings, confidence thresholds, activity signals, and public ranking. At least 30% of joins produce a later rating. Low rating volume makes rankings feel arbitrary.
3. Pull Community claims, moderation tools, saved communities, and weekly updates. 25% weekly returning explorers and organic operator submissions. Operators optimize listings without improving communities.

Decision: who can rate?

This changes both launch speed and the credibility of Top rated. Select a policy, then queue it for the next implementation pass.

Selection is local until queued.