Fricy

[RelayGPU @RelayGPU]

See MCP in action on Relay. ⚡

Connect your tools, models, and workflows through a simpler AI infrastructure.

Less friction. More control. One Relay.

Explore Relay: https://t.co/KVdFD8KBPN

[Fricy @Fricy274906]

@RelayGPU open-gpu:native

[RelayGPU @RelayGPU]

Google Veo 3.1 Fast is live on Relay. 🎬

Fast, high-quality video generation through the same API infrastructure you already use.

Build faster. Integrate simply. One Relay.

https://t.co/KVdFD8KBPN

[Fricy @Fricy274906]

@cortensor base:0xe30c0801a516f0c67ebc0c4b45ffa7dcbd72ea9a

[Cortensor @cortensor]

🛠️ DevLog – PyClaw: Latest Release Closes the Current External Bug-Report Batch

A quick follow-up on the recent external developer reports and the latest PyClaw release.

🔹 Reported issues resolved
- The current batch of externally reported defects has now been addressed and verified against the published release
- Fixes covered SDK tool exposure, complete tool-result handling, provider failures, deadlines, decorator behavior, approval handling, and MCP-related configuration
- The corresponding reports have been closed after release verification

🔹 Validation
- The fixes went through focused regression coverage across SDK, tools, providers, approvals, and runtime behavior
- Live provider checks were also included so the release was not validated only through deterministic tests

🔹 Why this matters
- This is exactly the external feedback loop we want to strengthen
- Developers use PyClaw, surface real issues, we triage them, ship fixes, and verify those fixes against the public package
- That makes the runtime more reliable for the next group of builders

🔹 What remains
- A few smaller SDK / configuration follow-ups are still tracked for later cleanup
- Those are already categorized and can move through the normal prioritization / release cycle

🔹 Current takeaway
- The latest release closes the current external bug-report batch
- More importantly, the public SDK and runtime are now starting to benefit from a repeatable external report → triage → fix → release → verification loop
- That feedback cycle should become even more useful as hackathon participation and broader developer usage increase

#Cortensor #DevLog #PyClaw #SDK #OpenSource #DeveloperTools #AgenticAI

[Fricy @Fricy274906]

@RelayGPU open-gpu:native

[RelayGPU @RelayGPU]

Relay is now agent-native. 🤖⚡

Official Relay MCP is live.

Connect Claude Code, Codex, Gemini and MCP-compatible agents directly to Relay.

npx @relaygpu/mcp

Your agent. Your models. One Relay.

https://t.co/VWiMMdG2Z5

[Fricy @Fricy274906]

@el_cano498 @cortensor @aixbt_agent @lukebelmar @VentureCoinist @DineroDom0 @game_for_one @ethermage @0xjesstech base:0xe30c0801a516f0c67ebc0c4b45ffa7dcbd72ea9a

[El_Cano @el_cano498]

✅ $COR @cortensor has spent nearly 2 years building. Mainnet Full Alpha, PyClaw 0.1, and real-world agent workflows are taking shape.

The tech is building. The market is still sleeping.

@aixbt_agent
@lukebelmar
@VentureCoinist
@DineroDom0
@game_for_one
@ethermage
@0xjesstech

[Fricy @Fricy274906]

@fatih_ogpu open-gpu:native

[Fatih @fatih_ogpu]

Every month the invoice comes back. Bigger. Scarier.

Your AI bill doesn't have to haunt you. One API. Every frontier model. Lower prices.

https://t.co/ke5sFlzyPe

Story, video and voice: 100% generated on Relay.

[Fricy @Fricy274906]

@Herbitualz @0xTP91 @virtuals_io @cortensor ethereum:0x8e0eef788350f40255d86dfe8d91ec0ad3a4547f

[HODLHenry @Herbitualz]

@0xTP91 @virtuals_io Joining @cortensor hackathon would be a massive step forward for @virtuals_io. Currently transitioning into mainnet

https://t.co/C0lCgAIpiX

[Fricy @Fricy274906]

@Kaffchad ethereum:0x8e0eef788350f40255d86dfe8d91ec0ad3a4547f

[Fricy @Fricy274906]

@cortensor base:0xe30c0801a516f0c67ebc0c4b45ffa7dcbd72ea9a

[Cortensor @cortensor]

🛠️ DevLog – Hackathon #5 Docs Are Getting Close to a Semi-Final Structure

We have refined the Hackathon #5 draft further, and the core structure is now getting close to a semi-final version.

🔹 What the round is centered on
- Build a small, working agent with PyClaw 0.1 + Cortensor Portal
- Keep the scope practical and reproducible rather than rewarding unnecessary complexity
- Use the round as both a builder competition and a developer-feedback exercise

🔹 Core requirements
- meaningful use of the PyClaw 0.1 Python SDK
- at least one real Portal-backed inference step
- a useful workflow involving tools, agent actions, or several connected steps
- a working demo that can be understood and reproduced

🔹 What we want to learn
- how quickly developers can get started with the SDK
- where examples, onboarding, errors, and documentation are still weak
- how Portal behaves under real agent workloads
- which tool-calling, structured-output, latency, or model edge cases appear in practice
- what kinds of useful agents builders naturally choose to create

🔹 Gemma4
- Gemma4 through Cortensor Portal remains an encouraged optional testing path
- Access will be coordinated because capacity is limited
- Useful Gemma4 integration and clear model feedback can receive additional consideration

🔹 Structure and judging
- The draft now includes clearer participation steps, submission requirements, demo expectations, judging criteria, qualification rules, and the general reward structure
- Working functionality, meaningful SDK usage, Portal integration, reproducibility, developer feedback, and usability are the main evaluation areas
- Qualification and placement remain organizer-reviewed rather than automatically guaranteed

🔹 Current status
- Timing, Portal access details, and the dedicated submission repository still need to be finalized before kickoff
- We will continue polishing wording and smaller details, but the overall builder flow, rules, evaluation structure, and intent of the round are now mostly in place

🔹 Current takeaway
- The goal remains simple: get builders using PyClaw + Portal, learn from real projects, and use that feedback to improve the SDK, docs, Portal integration, and overall developer experience before broader future rounds

Hackathon #5 draft:
https://t.co/7iKSq1HK0X

#Cortensor #DevLog #PyClaw #Portal #Hackathon #SDK #AgenticAI

[Fricy @Fricy274906]

@cortensor ethereum:0x8e0eef788350f40255d86dfe8d91ec0ad3a4547f SLEEPING GIANT.

[Cortensor @cortensor]

🛠️ DevLog – PyClaw: Hackathon #5 Docs and SDK Polish Become the Main Focus

For the rest of this week, a lot of the PyClaw work will shift toward Hackathon #5 preparation and tightening the builder experience around the 0.1 SDK.

🔹 Hackathon #5 docs
- The current draft is now centered on building small, working agents with PyClaw 0.1 + Cortensor Portal
- The goal is intentionally practical: useful, reproducible workflows rather than oversized demos
- Portal-backed inference, simple tools, multi-step workflows, and developer feedback are the core of the round :chatgpt-content-reference{index="0"}

🔹 SDK polish
- Keep tightening the install → define tool → create agent → run workflow path
- Improve examples, errors, setup, provider behavior, and release validation
- Make it easier for a hackathon builder to get from a fresh environment to a functioning tool-using agent without learning the full runtime first

🔹 Portal + Gemma testing
- Portal will be one of the main hosted inference paths for the hackathon
- Where capacity allows, Gemma4 testing can help expose tool-calling, structured-output, latency, and multi-step workflow behavior under real builder usage :chatgpt-content-reference{index="1"}

🔹 Why this matters
- Hackathon prep is also a product-quality exercise
- Better docs and SDK ergonomics should reduce onboarding friction before builders arrive
- Real submissions can then expose the next set of gaps across PyClaw, Portal, examples, model behavior, and developer UX

🔹 Current takeaway
- The near-term focus is less about adding another large runtime feature
- It is about making the current 0.1 baseline easier to understand, easier to build with, and easier to reproduce before Hackathon #5 kicks off

Hackathon #5:
https://t.co/7iKSq1HK0X

#Cortensor #DevLog #PyClaw #Portal #Hackathon #SDK #AgenticAI

[Fricy @Fricy274906]

@0xTwilty @arbitrum Crazy project ethereum:0x8e0eef788350f40255d86dfe8d91ec0ad3a4547f

[Twilty @0xTwilty]

You’ve been waiting. You’ve been patient.

$COR is entering Full Mainnet Alpha.

Built on $ARB - @arbitrum Layer 3 technology.