@getknos
Joined July 2025
  • Tweets29
  • Following7
  • Followers50
9Media
Join us for another edition of Sibyl Sundays. Talking about the new payments page, the MCP plugin and more. xitter.cf/i/spaces/1XxygwjMvobGM

8
19
1
40
1,637
Knos update The 1st version gave coding agents one shared memory so they stop stepping on each other. Then I ran into the next problem: agents keep saying “done” even when the tests fail. So Knos now pays an agent only when your tests actually pass.
3
5
101
Built on @sibyl_labs_ memory Why Solana? The money sits in a program, not with any company. It only moves when GitHub proves your tests actually ran. Still on devnet with test money for now. Code → github.com/drexthealpha/Knos Try it → drexthealpha.github.io/Knos/

GitHub - drexthealpha/Knos: The neutral meter for AI agent work: neither side keeps the count.

The neutral meter for AI agent work: neither side keeps the count. - drexthealpha/Knos

github.com
2
37
2
58
🤯🤯🤯🥳
Sibyl Labs
@sibyl_labs_Sep 16
Replying to @sibyl_labs_
Congratulations to our 1st Place winner @getknos 💰$4,000 + Network School Residency A shared memory for coding agents that enables experiential growth and unlocks new solutions by utilizing prior experience learned through usage. Big use case! xitter.cf/getknos/status/2098167… github.com/drexthealpha/Knos
8
4
32
923
What top agent memory systems need: 1. Sibyl: 95.6% accuracy, zero embeddings, local files 2. Typical vector DB: high cost, high hallucination risk 3. Independent 500-company test: Sibyl 100% retrieval Sibyl makes long-running agents possible.
Sibyl Labs
@sibyl_labs_Sep 13
Replying to @sibyl_labs_
Knos, by @getknos, gives coding agents a shared record of who is working on what. It can hold back answers and block conflicting edits when another agent has already claimed the work. xitter.cf/getknos/status/2098167… github.com/drexthealpha/Knos
1
2
13
773
Why Knos runs on Sibyl Memory: 1. Shared decision log across agents 2. Blocks conflicting edits automatically 3. 95.6% LongMemEval score (#2 overall) 4. File-based → no cloud, no model calls for recall Sibyl is the layer that makes multi-agent work reliable.
3
1
11
363
Knos, by @getknos, gives coding agents a shared record of who is working on what. It can hold back answers and block conflicting edits when another agent has already claimed the work. xitter.cf/getknos/status/2098167… github.com/drexthealpha/Knos
Knos@getknosSep 10
Who wants to see our @sibyl_labs_ hackathon entry? And for those who want to know the technical details of Knos and how Sibyl is its core:
2
2
1
14
1,366
💙
tradingtulips🌷
@tradingtulipsSep 12
base szn waiting room
1
4
269
Knos retweeted
Knos@getknosSep 10
Who wants to see our @sibyl_labs_ hackathon entry? And for those who want to know the technical details of Knos and how Sibyl is its core:
2
3
1
17
3,967
Who wants to see our @sibyl_labs_ hackathon entry? And for those who want to know the technical details of Knos and how Sibyl is its core:
2
3
1
17
3,967
Coding agents constantly step on each other. Unshared context slows down multi-agent innovation. A new solution is coming to change that. Thanks to @sibyl_labs_
2
14
268
more replies
Developers deserve better memory for their coding agents. Knos brings native, local coordination to your workflow. Stay tuned for reveal and launch details soon.
1
3
35
@sibyl_labs_
2
30
Meet Knos. One shared memory for every coding agent. Built on @sibyl_labs_ .
6
16
1
47
1,575
The first shared memory where coding agents claim work and refuse the rest. Context is not the only thing that can coordinate agents. @sibyl_labs_
2
3
21
360
Knos retweeted
ASML X@asml_on_xAug 21
ASML-X is live on @XLayerOfficial This is just v1 Get ready to plug in and test your own trading agents with hard limits and structural refusals Stay tuned
1
3
2
6
296
Load more