Useful things worth opening.

Isoquant

A new PAYGO inference provider launching with GLM-5.3-Flash at $0.07/$0.20 per 1M input/output tokens with automatic prompt caching; drop-in OpenAI-compatible API, no top-up fees. Its published benchmarks claim the lowest latency and highest throughput versus Together, CoreWeave, Baseten, Fireworks, and Z.ai.

⚠ Pricing and benchmark figures are vendor-reported and not independently verified.

Prism Inference

Pay-as-you-go inference with no minimums across serverless, elastic, private and dedicated endpoints, plus batch. Lists per-model rates (GLM-5.3 $1.40/$4.40, Kimi K3 $3/$15 per 1M tokens) and states strict zero data retention — inputs and outputs are never used for training.

InferHub

Proxy service that pools AI subscriptions into one API. Shared as "a rly interesting proxy" with per-plan pricing on its site.

AIHubMix

Multi-model API with per-model pricing, including a GLM 5.3 coding endpoint. Shared via its GLM 5.3 API pricing page.

Surplus Intelligence

OpenAI-compatible inference marketplace that routes each request to the cheapest current offer. Vendor claims ~38% savings via routing; pricing and retention terms weren't independently verified.

Kitani

Open inference on independent hardware with a prepaid USD balance — the max request cost is reserved, then settled on completion. API is OpenAI-compatible at api.kitani.ai/v1.

Pareto Inference

Pay-per-token inference running on Pareto's own GPUs, including GLM 5.3 Flash. Shared via its pricing page.

BytePlus ModelArk

BytePlus's Coding Plan for Claude Code, Cursor, Cline, Kilo Code, Roo Code and OpenCode, covering models like Dola-Seed-2.0-pro, GLM-5.1 and DeepSeek V4. Exact plan prices weren't visible on the promo page.

Parasail

Inference provider whose homepage FAQ states explicit zero data retention alongside SLA terms. Surfaced in #providers via a comparisons board post.

Inceptron

Inference provider with ISO 27001 certification, GDPR compliance and EU residency. No zero-retention statement was found on its site.

Subconscious

Coding plans starting at $100/mo with 60M tokens/day. Shared via its pricing page.

FEIHOA

Unlimited AI plans from €6. States prompts and outputs are not used for training, though it may retain data briefly for security, abuse prevention and legal reasons.

Crof

Inference provider at crof.ai; public evidence is thin (snippet-based).

⚠ Third-party substitution allegations seen during research are unverified and not repeated here.

Real API Pricing

GitHub repo behind the real-api-pricing comparison of actual inference API prices. Its web app is already in this library's web-apps section; this is the distinct source repo.

OpenCode Go

Subscription plan for the OpenCode coding agent. Shared with a screenshot of the plan page.

SingularityAPI

Inference API; a forwarded announcement noted DeepSeek-V4.1-Flash going live on it.

DeepSeek clock

Personal utility page showing DeepSeek's live peak/off-peak status in local time with both models' prices either way. Rates match the official docs.

Token Harbor

Inference API with a published pricing page; the sharer noted roughly 3.5x markup on its $10 plan.

StreamLake

Kuaishou's AI cloud / model-as-a-service platform with API and SDK integration for online inference. Holds ISO 27001/27017/42001 and MLPS Level 3; no explicit zero-retention statement found.

CoreWeave

GPU cloud infrastructure company offering serverless inference APIs, dedicated endpoints and GPU clusters. Infrastructure rather than a retail model API.

GMI Cloud

GPU cloud with serverless inference APIs and dedicated endpoints; lists H100 at $2/GPU-hour and H200 at $2.60/GPU-hour.

Venice

Private AI chat and API with free entry and paid tiers (Pro $18/mo). Its API carries a zero-data-retention promise.

Relace

Coding-focused API (Instant Apply model, code retrieval, embeddings, reranking) rather than general LLM inference. Available via Continue.dev, with a self-host/VPC option.

AntSeed

Open market for AI inference with no account or email required and optional TEE-backed providers. That design is not the same as a universal zero-retention promise.

Inco

Recently launched its own inference platform, announced on the company blog.

Kenari

Indonesian inference API — one API key, paid in Rupiah. Shared via its pricing page.

InferenceSaver

Comparison and routing service claiming 40–80% savings on AI inference costs.

Nube

Inference API with published per-token pricing. States explicit zero data retention with an upstream no-training clause.

Cline Pass

Subscription for Cline's coding agent at $9.99/mo, positioned as 2–5x the quota versus direct API spend. The "5x cheaper than API" phrasing seen in chatter is not verified.

Nebius Token Factory

Nebius's pay-per-token hosted inference for open text/vision models (~23 models incl. DeepSeek, Llama, Qwen). New accounts get trial credit; uses separate v1 API credentials.

Coral Bricks

High-throughput inference aimed at agents. A forwarded promo code (SUPERINT) covered GLM 5.2 and Kimi K3.

Fast Inference

Gateway for AI models and coding agents with provider failover. $9/mo gives $10 gateway credit, $49/mo gives $60, then pay-as-you-go at direct rates.

Entrim

$10 in credits plus pay-as-you-go inference. Processes prompts and outputs in RAM only and does not train on them.

Merge Gateway

Multi-provider gateway. A Discord-shared "75% off DeepSeek V4 Flash through September 30" promo was not corroborated on its site.

sference

Managed inference for open models on European GPUs.

⚠ Public evidence is snippet-based; details unverified.

Unifically

Pay-per-use multi-model API (no subscription) with a $0.20 free balance and per-generation pricing, e.g. $0.03 per image. The sharer called it "sus" — treat vendor claims cautiously.

Reflex

Post-trained Kimi K3 served with 6x context at a claimed 490 tokens/sec. Performance claims come from the provider's own page.

OpenModels

"The open market for AI tokens", with published plans.

OpenRouter

Multi-provider model gateway charging a 5.5% fee, with 50 free requests/day. Shared via its GLM 5.3 provider/pricing page.

TensorX

EU-hosted pay-as-you-go inference whose pricing page states verbatim: "all inference is zero data retention."

AI Router

Swiss-hosted inference API ("All of AI. One API. Made in Switzerland") with GDPR compliance and data hosted in Switzerland. Flat CHF 39/mo unlimited under a fair-use policy; OpenAI-compatible at api.airouter.ch/v1.

B.AI

Inference API documented with DeepSeek-V4-Flash endpoints.

OrcaRouter

Zero-markup BYOK router with an open-source self-hostable Lite (MIT); the hosted plan includes free credits. Pass-through per-token pricing, e.g. DeepSeek V4 Flash $0.15/$0.29 per 1M in/out.

OpenFerence

Inference API at openference.ai; public evidence is snippet-based only.

⚠ A Discord claim that its GLM 5.2 "is not even real" is unverified.

wafer

Inference provider, with DeepSeek V4 Flash mentioned on it in later chatter. Surfaced via an X post from @wafer_ai.

Fireworks AI

Serverless inference from $0.10/1M tokens (under 4B params), $1 free starter credit, 50% off batch. 100–400+ open models with an OpenAI-compatible API, FireAttention engine, LoRA/RFT fine-tuning and dedicated GPUs.

RunInfra

Inference provider; its DeepSeek V4 Flash $0.13/M input price was corroborated. A "1.8¢/M blended" figure seen in chatter is not a list price.

Freebuff

Free, ad-supported coding agent (CLI, desktop, web, cloud, chat) powered by open models like DeepSeek V4 Pro/Flash and MiniMax M3. Explicit no-training stance — it doesn't share data with third parties that would train on it.

GripHubRouter

Cheap pay-as-you-go multi-model router, reportedly built by the sharer's friend.

Melious

EU-hosted pay-as-you-go inference; the sharer cited ~240 tok/s on DeepSeek V4 Flash. States it does not train on user data.

SiliconFlow

One OpenAI-compatible API for open and commercial LLMs with serverless pay-as-you-go and dedicated endpoints. Homepage claim: "No data stored, ever."

CheaperInference

Pay-as-you-go inference with published example rates. Prompt and response bodies are not logged, with zero data retention on supported routes.

Novita

Inference provider, mentioned as a fan-out target in routing chatter.

Command Code

GOAT plan — $10/mo for $70 in credits (limits $14/5h, $35/week, $70/month) across 30+ open and closed models. No zero-retention statement found.

DeepSeek

DeepSeek's official API pricing: pay-as-you-go with 50% off-peak discounts (Flash $0.15/$0.60, Pro $0.66/$1.98 per 1M in/out on cache miss; peak is 2x). The $0.13/$0.30 figures seen in Discord are stale.

YOLO Auto

Unlimited-token API plans, currently $19/$39 — the Discord-era $6/mo claim is superseded. Shared as an unlimited Qwen3.6-35B API.

Claudin.io

Flat-rate API keys for coding agents, $9–$99.

regolo.ai

European inference platform with EUR per-token pay-as-you-go (e.g. Llama-3.3-70B €0.60/€2.70 per 1M in/out) and 70% off the first 3 months. Marketing claim: "Zero data retention on European green infrastructure."

Arli AI

Inference API with unlimited-request plans and an explicit zero-logging statement. The sharer noted it seemed new and unstable.

Hetzner

Hetzner's free experimental open-weight LLM inference API (announced July 2026), EU-hosted. Roughly 4M input / 100k output tokens per 60s per key, no SLA — not for production.

Synthetic

Inference provider; a forwarded announcement covered its scaling of Kimi K3.

Aster

OpenAI-compatible inference with free, Pro and Max tiers. States zero data retention on the free tier; Pro around $20 for $30 in tokens.

Featherless

Open-model inference provider. Only its status page was visible during research.

Meituan LongCat

Meituan's LongCat with Token Packs — 50M tokens for $4.90 (promotion corroborated). Surfaced via an X post.

DeepInfra

Pay-as-you-go inference across 100+ models with explicit zero data retention. Example: GLM-5.3 Flash $0.075/$0.25 per 1M.

CheapestInference

Flat-rate unlimited time-block subscriptions — Core pool from $22/mo per 8-hour daily block, Frontier $71/mo, Flagship $199/mo. Prompts and completions are processed in memory and discarded, never used for training.

PrivacyWatch

Open-source comparison of 86+ provider families (97 surfaces) on training, retention, zero-retention and data location, quoting each provider's own ToS and privacy docs. Maintained by the same crew as this library.

OpenAI API pricing

OpenAI's official per-1M-token pricing page (e.g. gpt-6-luna $0.10 in / $0.50 out short-context). Useful as the first-party reference behind third-party price comparisons.

Zro

$20/mo including $60 of inference spend, OpenAI/Anthropic-compatible. Explicit zero request retention — prompts and completions are never used for training, evaluation or analytics.

InferX

Pay-as-you-go inference across 200+ one-click production-ready model endpoints.

xKiro

Unified gateway — free tier with 40+ free models and 5M tokens/day (no card), pay-as-you-go about 35% below official prices on select models. 90+ models across 16 providers on one API key.

StepFun

StepFun's open platform with a usage-billed API (api.stepfun.ai/v1) and a 15-day trial needing no card. Serves Step 5 Preview and StepAudio 3.

Phoenix Grove

Coding-plan provider; the sharer noted a vague allowance but a free month on offer.

Neuralwatt

$10/kWh pay-as-you-go (or per-token), $1 free credit after adding a payment method; monthly plans $20–$200. Hosted multi-model OpenAI-compatible API plus on-prem deployment; 14 models.

Chutes

Inference provider that was notably early serving Kimi K3, per channel chatter.

Telnyx

Inference API with published per-token pricing — Kimi K3 at $2.70/M input, $0.27/M cached input, $13.50/M output. The sharer noted they're open to negotiating volume pricing.

Charm Hyper

Free monthly credits plus a $20/mo plan, with an explicit zero-data-retention and GDPR claim.

TokenWatch

Live price comparison across inference providers, maintained by the same crew as this library. The sharer noted a Hyper subscription yields $12.50 of inference a day.

Zyloo

Subscription-based inference provider at zyloo.io.

Moonshot platform docs

Moonshot AI's platform documentation for the Kimi model family. Shared originally via a Kimi-K3 Hugging Face link.

Verboo Code

Free trial, then unlimited-token subscriptions from Junior $24/mo to Ultra $269/mo, plus a usage API from $0.12/1M input tokens. Coding agent plus OpenAI-compatible API on Brazilian servers; states it does not train on or review code and calls.

Makora

Inference provider and optimization service — Starter $20/mo and Developer $200/mo (both shown sold out), with private model serving and on-prem options. Starter covers unlimited models under 40B parameters.

Railway Changelog #0309: Free VMs without an account

Railway Changelog #0309 opens free Linux VMs with no account: `ssh railway.new` gets a 2 vCPU/2 GB box with coding agents preinstalled, 60 minutes to build and 24 hours to claim it before deletion. The same edition adds request tracing in Priority Boarding (edge-to-database waterfalls with no code changes) and OpenCode on Cloud Agents for persistent agent VMs.

Orca

A free, open-source Agent Development Environment (ADE): run Claude Code, Codex, Gemini, Cursor CLI, and other CLI agents in parallel across isolated git worktrees, with a Ghostty-class terminal, in-app diff review, embedded browser, and SSH dev built in. Bring your own subscriptions — agents run side by side without interfering.

heddlework

A native, harness-neutral workspace for agent sessions: task graphs, diffs, and durable work across coding agents — Pi first, Codex and Claude next.

⚠ Very early (~12 stars); expect rough edges.

gpui-kit

Rust GUI components for building cross-platform desktop apps on GPUI (the framework behind Zed). ~14.8k stars — one of the more-starred GPUI component libraries.

scenario-labs/skills

Agent skills for Scenario: image, video, audio, texture, 3D, and custom-model workflows over the Scenario MCP, aimed at games, entertainment, and creative verticals. Installs with `npx skills add scenario-labs/skills`; ~450 stars.

Detail

A curated gallery of the small UI decisions that make products feel considered — smarter word breaks, corner-morph animations, name-derived placeholders — each shown as a short clip. Curated by Rene Wang; a good reference for stealing micro-interaction ideas.

Landbook

A website design inspiration gallery with a large curated collection of landing pages, e-commerce sites, portfolios, and templates, plus boards for saving references. Useful for browsing real-world site design by category.

MaxiBestOf

A website design and typography inspiration feed updated every weekday, pairing featured sites with their font stacks (e.g., Geist, Neue Haas Grotesk). A Pro tier adds private collections and HD screenshots.

SoulStack

ProxySoul × Empryo: a bundle of skills, agents, an "immune system," and habits for coding agents.

⚠ Very early (~17 stars); concept-heavy and expect rough edges.

GLiNER2.5-Decide

Fastino's 340M-parameter open-weight model for schema-defined decision-making: feed it text plus typed questions and get valid answers with probabilities, confidence scores, and constraint-feasibility metadata. Apache 2.0, CPU-capable, strong on routing/triage/guardrail tasks.

⚠ Benchmark figures are vendor-reported and not independently verified.

pi-warden

Guardrails for Pi built on pi-typesafe — "steer the agent instead of interrupting you: Jev judges." Part of a small Pi-safety ecosystem from the same author as the next entry.

pi-typesafe

TypeSafe decisions for Pi — a batched evaluation tool, terminal playground and typed API for extensions. The foundation pi-warden builds on.

BrokenArXiv / ArXivMath (announcement)

Jasper Dekoninck releasing new versions of the BrokenArXiv and ArXivMath benchmarks.

⚠ Announcement only; useful pointer for anyone tracking eval benchmarks.

Arrow 2 (announcement)

QuiverAI's Arrow 2 — their latest models for generating precise, editable vector graphics.

⚠ Announcement only; a notable data point in AI vector-graphics generation.

MiniCPM5-2B-PCD

MiniCPM5 2B PCD weights on Hugging Face — the same model family as the 16 Sept in-browser Pi announcement. The model behind that WebGPU coding-agent demo.

pi-claude-agent-sdk

An inference provider that lets Pi (a terminal agent harness) run on Claude Code Pro/Max subscriptions instead of API keys.

⚠ Kenn is using a fork and admits he "just closed my eyes and let the agent figure it out" — expect rough edges.

pi-claude-bridge

Lets you route Pi's model calls through a Claude subscription ("allow sub usage in pi"). Same cost-saving idea as pi-claude-agent-sdk: agentic coding without per-token API billing.

spirula-studio

Cross-vendor 3D Gaussian Splatting trainer — video to splat to mesh, running on Vulkan or CUDA. Notable for not being CUDA-locked.

Responsively App

A dedicated browser for responsive web development — preview a site at multiple screen sizes side by side. Built for faster, more precise responsive work.

Grokbot Characters (Rege Lab)

A Rege Lab page with six little Grokbot character expressions to play with, plus downloadable editable CoAnimato files. A creative asset resource for bot and character design.

Getting the most out of Opus 5.5

Official Claude guide to prompting Opus 5.5, steering long agent runs, and checking results in Claude apps and Claude Code. A practical reference for anyone driving long agent sessions.

Grokbot Icon Studio

A minimal bot-icon studio — browse bot icon examples and copy a prompt to generate an icon of your own. Quick asset generation for bot projects.

Cloudflare Quick Tunnels

Cloudflare Quick Tunnels — free encrypted tunnels that expose localhost as a public URL. Cata plans to use it instead of the VS Code tunnel.

BendSR

Symbolic regression implemented in Bend. A research-flavoured repo for anyone exploring the Bend language.

security-audit-skill

A coding-agent skill for multi-phase security audits with independently verified, machine-readable findings. A drop-in skill for agentic security reviews.

Saving 100TB of RAM with math

Cloudflare engineering deep-dive on saving 100TB of RAM with math (and Rust). A solid systems-engineering read.

Bend

Bend 2 — a fast programming language that blocks AI mistakes via proof checking. Install with curl from bend-lang.com.

Monid free agent search (announcement)

Announcement of a free agent search/fetch API — "killed Exa, Tavily and Brave", $0 with no quotas, from Tiny_Fish and MonidHQ.

⚠ The pricing claims are the poster's and unverified; this link is the announcement, not documentation.

jev-search

Shared by thelaggingway with just "Another one" — context is thin, but it landed in the search-tool thread. Worth a look if you're tracking search APIs.

ArgusAgent

Microsoft's ArgusAgent — a persistent, reviewed multi-agent runtime for long-horizon research and engineering. Directly relevant to anyone running agents on long tasks.

PrettyMD

A markdown viewer "for vibe coders" — a simple, beautiful app for reading the .md files your AI writes: specs, plans, skills, agents and more. Pairs naturally with agent-heavy workflows.

LFM2.5-2.6B-RLCD

LiquidAI's LFM2.5 2.6B model tuned for RLCD, shared in the context of Jev's API and concepts. A small local model for anyone experimenting with RLCD-style agent loops.

Qwen-2.5-1B-RLCD

A 1B RLCD-tuned Qwen 2.5 that kenn tried locally "to get a feel of how Jev's API and concepts might feel like" — and found it did fine in a roguelike spike. Useful as a tiny testbed model before spending on bigger runs.

Pi coding agent in the browser (announcement)

Announcement of a coding agent (Pi) running entirely in the browser on a 2B model via WebGPU — MiniCPM5-2B + Pi with Transformers.js and 4-bit ONNX weights.

⚠ Announcement only; no repo was linked in the post.

varve

An experimental Rust time-series database with DuckDB SQL and tiered Parquet storage — "time settles into layers." Early-stage but interesting for anyone into time-series storage design.

ever-gauzy

Ever Gauzy — an open business management platform covering ERP, CRM, HRM, ATS and project management (gauzy.co). A full open-source suite if you ever need to stand up business tooling.

Universal syntax highlighter / OpaqueRange (demo)

Demo of a universal syntax highlighter built on the new OpaqueRange web API (Chrome 152+, 2026) — syntax-highlighting different program fragments efficiently. Worth a look for anyone building code-rendering UI.

The invisible side of design engineering

Jakub Krehel's essay on the invisible side of design engineering — the craft that doesn't show up in the final pixels. A thoughtful read for design-minded engineers.

Inspo

A reference layer for AI coding agents — real production websites with desktop and mobile captures, "real websites worth studying." Useful as grounding material when agents need to see how good sites are actually built.

FeralUI

A small library of playful, physics-driven React elements — including a ceiling pull-cord with real Verlet rope physics. For interfaces that should feel alive rather than flat.

GVS5H

A harness claiming five Qwen3.8-27B models match Claude Fable 5 on LiveCodeBench Hard — "Fable 5 level coding." The performance claims are the poster's and unverified; scarywood75 noted the implementation is Python.

transitions.dev

A collection of the most essential transitions for web apps, plus a skill for agents and a Refine tool. Drop-in motion craft for web projects and the agents building them.

UI Skills

A collection of design-engineering skills covering accessibility, motion and frontend craft. Pairs well with the transitions.dev repo above for leveling up UI output.

e2e agent testing framework (announcement)

Oskar's announcement of "e2e", a new agent-oriented testing framework — he ported React Native Pager View's test suite from Maestro to it.

⚠ Announcement only; the framework itself is still being built.

pdfcn

A collection of beautifully designed, accessible and customizable PDF components built on Takumi. Handy if you ever need to render polished PDFs from a web stack.

pascalorg/editor

An open-source 3D architectural editor with a local CLI, MCP tools and practical workflows for humans and agents. Niche but complete — 3D editing with agent integration baked in.

Brilliant

A professional vector design tool "where you and your AI agents work directly on the canvas." Agent-native design tooling is still rare, which makes this one notable.

Real API Pricing

Compares the real cost of AI — subscription allowances, model benchmarks and the price–performance frontier, with transparent sources. Useful before committing spend to any provider.

pi-arm64-exec (announcement)

A pi extension giving the agent exactly one tool — doing file read/edit work in arm64 assembly — to stop Astra shelling out to python3, ruby and perl for file ops. Announcement; the actual link was promised in a follow-up tweet.

Minara Harness (announcement)

Announcement of Minara Harness — pitched as "general financial intelligence for everyone", turning ideas into signals into execution, plus coding tasks like any other agent.

⚠ The capability claims are the poster's and unverified.

herdr fleet management update (announcement)

Daniel's post that a new herdrdev update is out with fleet management live. A product-update announcement; the herdr blog below has the substance.

How to Build 3D Worlds with Astra

A step-by-step guide to making photorealistic 3D worlds and games — reference photos, Blender assets, sub-agents, blind critics, Three.js vs Unreal Engine, and a starter prompt. The "GPT-6 Astra" framing is the site's own marketing.

OSIRIS

An open-source Palantir alternative — track 10K+ aircraft, 2K satellites and worldwide CCTV on a 3D globe, and run Nmap, DNS, WHOIS and threat-intel scans from the browser across 20+ live feeds. Free.

open-pencil

An AI-native design editor pitched as an open-source Figma alternative. Worth watching if design tooling is on your radar.

Herdr 0.9 — Connecting the machines

The Herdr blog on 0.9, which brings local and remote machines into one TUI client, with Herdr Cloud coming next. Herdr's pitch: run agents anywhere, leave them running.

ECC

An agent-harness performance optimization system — skills, instincts, memory, security and research-first development for Claude Code, Codex, Opencode, Cursor and beyond. Directly relevant if you're tuning agent setups.

MarkItDown (announcement)

Matt Dancho's post on Microsoft's MarkItDown — turns almost any file (PDFs, Word, PowerPoint, Excel, images) into clean Markdown; claimed 100,000+ GitHub stars.

⚠ Announcement only; the repo wasn't directly linked.

Magika (announcement)

Google's Magika — detects the real content of any file even when disguised (malware posing as PDF, scripts hidden as images); free and open-source.

⚠ Announcement only; the repo wasn't directly linked.

teamai-cli

Tencent's "Make Every Team AI Native" CLI. The description is thin, but a Tencent-backed team-AI CLI is worth a look.

pi-mockdeck

Cata's first pi extension — an ASCII showcase/mockup gallery of UIs for fast iteration. The shared link pointed at the commits page; normalized to the repo root.

okf-agent-memory

Git-native persistent memory for AI coding agents — implements Google OKF v0.2 with sub-300µs in-memory BM25 search and an embedded MCP server.

⚠ Claims to slash token bloat by 80%; that figure is the repo's and unverified.

tgrep (announcement)

Microsoft's tgrep — grep for large codebases with client/server support, claimed 7–50x faster than ripgrep and ugrep.

⚠ Announcement only; the speedup claim is the poster's.

mobframework

"BEAM on device" — write mobile apps in Elixir with the whole BEAM running on the device, rendering native iOS and Android UI; your coding agent attaches to the running app with real state and hot-pushed fixes. Ambitious and unusual.

ripwire

"The ripgrep of AI context" — a zero-dependency C++23 CLI plus MCP server that gives coding agents a ranked, deterministic map of any repo: blast radius, tests-to-run and quality deltas. From Red Hat's emerging-tech group.

Architectural visualization with Astra

OpenAI's developer blog on using Astra to design a house, refine everyday details, and explore it through Blender camera tours and an Unreal Engine 5 walkthrough. A concrete agentic-design case study.

SponsorBar (announcement)

SponsorBar — puts a small ad in your menu bar and shares the revenue with you ("make money from your menu bar for doing nothing"), with a 10%-extra launch promo. A real launched product, though a niche one.

Factory Open Source AutoWikis

Browse Factory-generated wikis for popular open-source repositories — a preview of what living documentation looks like on real codebases before running it on your own. Nice for evaluating doc-generation tooling.

monocode

A GUI for your coding agents. Description is minimal, but a visual front-end for agent harnesses is a gap worth filling.

synthetic-search-mcp

A search MCP server from baanish, who vouched for PR #4 personally. The shared link was the PR; normalized to the repo root.

Psychology for UX: Study Guide

Nielsen Norman Group's study guide on the psychology behind UX. A solid reference for grounding design decisions in research rather than taste.

beUI motion components

Copy-paste animated React components from beUI. Drop-in motion without building animation primitives from scratch.

Canvas UI

A creative component library for Canvas, WebGL and WebGPU. For interfaces that need GPU-level visuals rather than DOM elements.

treg

A tool catalog for agents — 2,630 endpoints across 47 providers (SEO, social, enrichment, ads data) callable with one key, priced per call, plus support for your own keys and skills. Tom's 2 Sep share carried his referral param; normalized to the root domain.

Claude for people-search (announcement)

Jason Zhou's announcement of "Claude for people-search" at a claimed $0.0089/lead versus $600 subscriptions.

⚠ The pricing claims are the poster's and unverified; the tweet mentions a git repo "below" that wasn't captured.

Paper

Paper — "design, share, ship." A design tool baanish was scouting; no verdict from the channel yet. (Rykuuun's pricing-page share from 8 Sep was dropped as context-free.)

Paseo

Kenn wanted to fork Paseo for internal use cases but held off because it was AGPL — PR #3944 has now relicensed it to Apache. The shared link was the PR; normalized to the repo root. The license change is the news: it's forkable now.

elxr

asx8678's own repo, built after burning through tokens fast. Context is thin, but it reads as a token-efficiency tool from someone feeling the pain directly.

DSH — DeepSeek Harness (announcement)

The Elixir Forum announcing DSH, the DeepSeek Harness.

⚠ Announcement only; no repo or docs link was captured.

Fresh

Fresh — "The Terminal IDE." A terminal-based IDE worth a look if you live in the shell.

Markdown Graphs

Monospace charts and tables for MDX and React. Handy for rendering data visuals that stay crisp in markdown contexts.

Eclipse Theia

Eclipse Theia — a cloud and desktop IDE framework implemented in TypeScript. The open foundation behind several well-known IDE products.

Plane

Open-source alternative to Jira, Linear, Monday and ClickUp. A self-hostable project-management option for teams avoiding SaaS lock-in.

Hy4-preview-FP8

Tencent's Hy4 preview weights in FP8 on Hugging Face — 770B total, 49B active, 1M context, open weights. For anyone with the hardware to run a frontier-scale open model.

GLM-5.3-Flash-DFlash2

A GLM-5.3-Flash quant on Hugging Face. Context is thin — thelaggingway's whole review was "Huh" — but it's another fast-model option to know about.

DBX

DBX — a 20 MB tool for managing 90+ databases. Remarkably small footprint for a universal database client.

AoAgents

An AI agent platform whose pitch is "Everything is an Agent".

⚠ Announcement-level description only; details unverified since the link was not opened.

Phoenix (Arize AI)

Arize AI's open-source observability platform for LLM apps — tracing, evaluation, and experimentation for agent and RAG workloads. A standard choice for watching what production agents actually do.

Prime Agent (Prime Intellect)

Prime Intellect's framework for training AI agents with reinforcement learning.

⚠ Per the sharer, its gimmick is collecting training data through its CLI — worth a look if you're into RL-for-agents; specifics unverified.

windows-harness (browser-use)

A Windows automation harness from the browser-use team (the web-agent library folks).

⚠ Likely for running agents against Windows apps; details unverified.

Spec-PTC: Speculative Parallel Test-Time Compute

A write-up of Spec-PTC, a speculative parallel test-time compute framework for long-horizon agents. Interesting if you follow inference-scaling tricks for agentic workloads.

Headlong (Laude Institute)

An open-source agent microharness with persistent agency and recursive LLMs — "of bash, by bash". Minimal by design; good for experimenting with agent loops without a heavy framework.

Shading Motion — Maxime Heckel

Maxime Heckel's deep dive into shading-based motion design techniques. One of the better-written front-end animation articles; useful if you care about motion craft.

Radius — The AI Browser for Research

An AI-native browser aimed at research workflows.

⚠ Product claims unverified; notable as another entry in the agentic-browser wave.

darkbloom — MCP gateway

An MCP (Model Context Protocol) gateway that advertises an accuracy guarantee — unusual positioning in the MCP tooling space.

⚠ Claims unverified; interesting if you run MCP servers in production.

pi harness v2 docs (earendil-works)

Deep link into the pi coding agent's v2 harness documentation. Worth reading if you use pi or are designing your own agent harness.

ThreeUI

A browsable library of Three.js components and interactive shaders. Handy for dropping 3D effects into web projects without writing raw WebGL.

kolu

A terminal tool that bills itself as "the best way to run terminals".

⚠ Details unverified, but worth a look if your terminal setup needs rethinking.

anti-slop (dmmulroy)

Opinionated Oxlint rules that reject low-evidence TypeScript/JavaScript patterns — basically a linter config against AI-slop code. Fun and practical for teams drowning in generated code.

Introducing Slack Code (Salesforce)

Salesforce's announcement of Slack Code, bringing coding-agent workflows into Slack. Announcement-level; the "oh no" from the sharer suggests healthy skepticism.

claudish-to-english (gvzdv)

A small repo that translates "Claudish" — AI-assistant jargon — into plain English. Niche but amusing; useful if you've ever stared at a model hedging for three paragraphs.

interior.dev

Micro-interactions for React, "built for the half-second after a click". A focused little library for the small animations that make UIs feel alive.

eve — durable AI agent framework

A durable agent framework pitched as "Next.js for agents" — build agents with one folder, with durability baked in.

⚠ Product claims unverified; the durability angle is the interesting bit.

OpenHistory

Turns your Mac activity into a private timeline for you and your AI — local activity history as agent context.

⚠ Privacy-sensitive by nature; the "private" claim is unverified.

Berd

A self-described "weird, playful" desktop app for building with AI agents — "less chatting, more building".

⚠ Product claims unverified.

fx — Vercel Labs coding agent

Vercel's announcement of fx, "a tiny, open, native coding agent" from Vercel Labs, originally an internal tool.

⚠ Announcement-only source; details unverified.

skills (emilkowalski)

A collection of "skills" for designers and engineers from Emil Kowalski (of Vercel/animations fame). Likely agent-skill style playbooks; worth browsing.

GPS-01 — interactive GPS explainer

A tiny interactive explainer of how your phone finds itself — satellites, signal timing, and trilateration. The good kind of technical storytelling.

Rare UI

A library of animated React components. Another solid option in the crowded React-animation space; compare with the other UI libraries already in the archive.

Cua Computer History (open source)

Cua's announcement of the first open-source "Computer History", in early preview. Announcement-only; interesting for computer-use agent research.

watermarks-remover (guillaumemeyer)

Strips multi-vendor AI provenance marks — Unicode text hygiene, statistical rewrite hooks, and C2PA handling. Legitimate hygiene tooling; obviously dual-use.

codedb (justrach)

A Zig code-intelligence server and MCP toolset for AI agents, recommended firsthand by the sharer. Native-speed code intelligence as agent context is a strong pitch.

codebase-memory-mcp (DeusData)

Persistent semantic memory for coding agents — an MCP server that indexes your codebase. Useful when you have many repos and context windows aren't enough.

quickbeam (elixir-volt)

A JavaScript runtime for the BEAM — Web APIs backed by OTP, native DOM, and a built-in TypeScript toolchain. Genuinely novel: Erlang's concurrency model meets JS.

Qwen3.6-27B-MTP-GGUF (unsloth quant)

Unsloth's GGUF quantization of Qwen3.6-27B with MTP (multi-token prediction). For local inference of the 27B Qwen.

Qwen3.8 open weights announcement

Alibaba's announcement that Qwen3.8's open weights are out. Announcement-only; pair with the unsloth GGUF quant below for local runs.

Qwen3.8-27B-GGUF (unsloth quant)

Unsloth's GGUF quantization of Qwen3.8-27B for local inference. The practical artifact behind the open-weights announcement above.

pi-acp-jetbrain (ryan-brosas)

A JetBrains IntelliJ ACP (Agent Client Protocol) adapter for the pi coding agent. Niche but exactly what you need if you want pi inside IntelliJ.

Woxi (vs Mathematica comparison)

Woxi's comparison of itself against Mathematica — Woxi being an interpreter for a subset of the Wolfram Language. A decent way into what Woxi is and where it stands.

spec-kit (GitHub)

GitHub's official toolkit for spec-driven development — write the spec, let agents implement. From GitHub itself, so it carries weight in the SDD workflow trend.

pullfrog

An open-source, model-agnostic, BYOK GitHub bot that runs in GitHub Actions. Bring your own key, pick your model, automate your repo — the self-hosted answer to proprietary coding agents.

DeepSeek Harness (developer preview)

DeepSeek's agent harness in developer preview — "everything is a plugin", with source released alongside. Every agent capability is a swappable plugin; the plugin architecture is the interesting idea here.

⚠ Claims unverified.

Delta — multiplayer coding with agents

A multiplayer environment for coding with agents — shared threads where teammates and agents build and review together, with full conversation and code context.

⚠ Product claims unverified.

Rust Foundation Trusted Training (RFTT)

The Rust Foundation's quality standard for Rust training. Reference material if you're evaluating Rust courses or building one.

hax

A minimalist coding agent that runs natively in the terminal.

⚠ The "minimalist" claim comes from the sharer's one-line description and is unverified.

AgentFS introduction (Turso docs)

Turso's introduction to AgentFS, which Rykuuun found interesting.

⚠ It appears to be an agent-oriented filesystem from Turso; anything beyond the docs link is unverified.

billion-context-pi

Rykuuun shared this asking "this is pi-vcc right?", suggesting it's related to a "pi" coding agent with very large context.

⚠ Details are unverified from the message text alone.

pi (nktkt)

A GitHub repo named "pi", shared without description.

⚠ Given the surrounding chatter about pi-vcc, likely an agent project; unverified from the message text.

maki

An AI coding agent (Rust, ratatui TUI) optimized for minimal context-token use: a tree-sitter `index` tool summarizes file structure before reading, system prompts stay concise, and subagents can pick weak/medium/strong models per task. It is extendable via neovim-like Lua plugins, supports skills/MCPs, session memory, and plan mode, and works as an agent backend for editors like Zed via ACP; MIT-licensed. Its benchmark claims are the author's own.

Inspora

A design-focused site, shared link-only.

⚠ Its exact purpose is unverified from the message text; the domain suggests a design inspiration or resource site.

Gooey

A site by Jakub Antalik, shared link-only.

⚠ The name suggests gooey/morphing visual effects; unverified from the message text.

Collect UI

A gallery of collected user interfaces for design inspiration.

⚠ Shared link-only; the description is inferred from the site's name and unverified from the message.

bro-skill

A one-file Agent Skills-format skill adding a `/bro` command: when a coding agent's last reply was too dense or jargon-heavy, `/bro` re-explains that same answer in plain language, with facts, paths, and commands preserved verbatim. It ships an installer covering Hermes, Claude Code, Cursor, Codex, and opencode; MIT-licensed.

Avoid z-index Whenever Possible

A 2020 article by Daniel Rotter arguing for CSS layouts that avoid z-index.

⚠ The thesis is in the URL slug; the article's quality is unverified.

Beautiful UI

A demo site showcasing a "beautiful UI".

⚠ Shared link-only; what it demonstrates is unverified.

oil-motion

An agent-oriented skill (SKILL.md) for designing and implementing web animations that respond to scroll, drag, pointer, touch, or state changes. The agent confirms keyframes, generates in-between motion with AI video models, then converts it into controllable web assets (MP4, alpha WebP sprite sheets) with quality checks for frame consistency and mobile performance; MIT-licensed. The README is primarily in Chinese (an English README.en.md exists), and the generation step needs external AI video services and API keys.

Arlan Marat's Vault

A public vault of UI/design techniques by designer Arlan Marat — dated entries like "Arcade pixel," "Holo," "Fade motion," "Liquid UI," "Kinetic typography," and "Apple's corners," each demonstrating a copyable visual effect. Everything is MIT-licensed and explicitly free to copy.

Refero Design Styles

A library of 2,000+ "AI-readable" design systems harvested from leading product websites (Apple, Cursor, Linear, Vercel, ElevenLabs). Each style exposes colors, typography, spacing, and components plus a DESIGN.md file meant to be fed into Cursor, Claude Code, Codex, v0, or Lovable to give AI builders better design taste; currently in beta with new styles added weekly.

There Is An Open Source For That

A directory of open-source alternatives to popular software.

⚠ The description is inferred from the site's name; unverified from the message.

josh

"Just One Single History" — a git-at-scale platform built on fast, reversible Git history transformation. It presents any part of a repo (e.g. a monorepo subfolder) as an independent repository with history preserved, so repo boundaries can be decided per-context instead of reflecting org silos; it ships a `josh` CLI for partial repo projections and `josh-proxy`, a Git HTTP/SSH proxy with a GraphQL API for CI/build queries. A mature project (since 2016, MIT), though some listed features like the merge queue and Starlark filters are not fully mature yet.

Checklist Design

An AI-powered design review tool: run a UX/UI quality check on any screen and get instant analysis of what's working, what's missing, and what to improve. It also hosts a searchable library of design checklists covering websites, web apps, mobile, components, and flows, plus a Figma plugin for in-canvas reviews.

oh-my-hermes

asx8678's entire review was "crazy".

⚠ What the repo does is unverified from the message text.

Kitesurf (Cloudflare blog)

A Cloudflare blog post about Kitesurf.

⚠ The post's claims are unverified; it's a vendor blog.

linus-torvalds-skill

Distills Linus Torvalds' code-review methodology from ~38k review moves extracted from 31k of his LKML emails (2002–2026) plus 67 interview transcripts into a reusable, language-agnostic agent skill. Ships four variants distilled by different LLMs (gpt-oss, GLM, Mistral, Qwen) with validation reports comparing skilled vs. baseline reviews; the gpt-oss variant is the recommended default.

bb

An agentic IDE that "builds itself" — it can control, customize, and automate itself, pitched as the groundwork for your own software factory. Desktop app, web app, CLI, and HTTP API are all first-class surfaces; work runs in threads you can follow live, steer at any point, or hand off to another agent. In active development; core architecture is stable, workflows and surfaces still evolving.

Qwen3-8 Max on Artificial Analysis

Artificial Analysis's model page for Qwen3-8 Max, with benchmarks and pricing.

⚠ Shared link-only; the page's data is third-party and unverified.

opencli

Turns any website into a CLI an AI agent can use: opencli drives your logged-in browser session (Playwright-based) so agents can operate sites through a command line instead of fragile DOM scripting. ~30k stars.

desengs

Shared link-only.

⚠ Purpose unverified from the message text; the domain suggests a design/engineering resource.

Antidoom (Liquid AI blog)

A Liquid AI blog post about "Antidoom".

⚠ The post's claims are unverified; it's a vendor blog.

Kimi K3 on MI355X (Wafer blog)

A Wafer blog post benchmarking Kimi K3 on AMD MI355X hardware.

⚠ Claims unverified; vendor blog.

Snowflow

The live demo of the snowflow_demo repo below: real-time procedural snow rendering in the browser via WebGPU, Babylon.js, and hand-written WGSL — GPU-generated terrain, snow deformation, procedural characters, cloth, and post-processing with zero textures, meshes, HDRIs, or animation assets. Requires a WebGPU-capable browser (Chrome 113+ on desktop GPU) to load.

snowflow_demo

Source repo for the Snowflow demo above: a fully procedural real-time snow scene built on WebGPU + Babylon.js with hand-written WGSL shaders — terrain, snow deformation, surf wakes, water spells, atmosphere, and post-processing, all generated at runtime. ~580 stars.

IRIS

IRIS (Intelligent Real-time Input Suggestion) is an open-source shell auto-completion tool positioned as the successor to Fig: a single native Go binary that runs purely on TTY — no Electron, GUI, account, or telemetry — rendering suggestion menus inline in the real terminal, over SSH and inside tmux. It indexes your aliases and shell history, supports AI suggestions via API key or local models, and configures via a TOML file with instant reload. No Windows support, and it can conflict with other autosuggestion plugins (zsh-autosuggestions, atuin), which the README says to disable first.

kimi-k3-mlx

An MLX port of moonshotai's Kimi-K3 (2.78T multimodal MoE); the weights are converted from the original release and requantized.

⚠ Conversion quality is unverified.

Hermes Console

A self-hosted AI agent console ("Tu agente AI self-hosted") that thelaggingway is giving a go.

⚠ The self-hosted claim is unverified beyond the site's tagline.

OptMem

Permanent memory for AI agents by Victor Taelin: the whole integration is a 426-token prompt plus a single dependency-free Python script installed with one curl command. You paste a `## Memory` block into your agent's AGENTS.md/CLAUDE.md, then use `memo wake` (read memory at session start) and `memo note` (record one-liners), plus `nap`, `recall`, `zoom`, and `forget`; storage is an append-only log with a rebuildable binary-tree summary cache (1M memories ≈ 608 MB, wake in 0.03s).

Content Architecture

A commercial, agent-ready Sanity CMS starter kit in Next.js 16 and Astro 7 editions: a page-builder CMS with in-app Studio, visual editing, draft mode, SEO helpers, a seed dataset, and Plop scaffolding. Its differentiator is agentic-coding workflow support — it ships AGENTS.md plus scoped skills, two preconfigured MCP servers (one reads the running dev server, one drives real Chrome for screenshots/traces), an editable llms.txt, and token-light Markdown versions of every page served to agents via content negotiation. One license per buyer covers unlimited personal, commercial, and client projects; it is a paid product, not open source.

nimbus-lite

A free, framework-free AI chat UI kit in vanilla HTML/CSS/JS — open index.html and it runs, no build step or dependencies. It includes a full chat layout with sidebar, live token streaming with blinking caret, Markdown plus syntax-highlighted code blocks with copy buttons, a model picker, and light/dark themes with accent colors; it is front-end only, so you bring your own backend (OpenAI, Anthropic, or a local model). The maintainer sells a fuller Nimbus kit ($19) adding citations, reasoning blocks, tool-call chips, and attachments.

worldmonitor

An open-source (AGPL-3.0) real-time global intelligence dashboard: AI-synthesized news briefs, a 3D globe and WebGL flat map, a live Country Instability Index for 31 Tier-1 countries, and cross-stream correlation of military, economic, and disaster signals. It is built for programmatic access too — MCP server, REST API with OpenAPI spec, CLI (`npx worldmonitor`), and Python/Ruby/Go SDKs, with local-AI support via Ollama — and ships as a web app, Tauri desktop binaries, and PWA. AGPL-3.0 copyleft applies, so commercial/SaaS use needs a separate license or AGPL compliance.

Amicro

A curated gallery of premium React micro-interaction and transition components built with React and Motion, with layout inspired by transition.dev. It showcases around 36 named interactions — morph, glare shine, magnetic field, text reveal, focus blur, shake, rotate, color-morph buttons — as ready-to-use UI components. Licensing and copy-paste terms are not stated on the page, so reusability is uncertain.

pi-dispatch

Runs the pi coding agent as a self-hosted background service: triggered on demand, on a cron schedule, or by GitHub/GitLab issues, comments, and pull/merge requests, each job executing in a locked-down ephemeral container with a durable queue. It enforces per-job/daily/weekly/monthly spend caps before a single token is spent, and ships a live admin panel with run history, cost insights, and per-trigger spend analytics; MIT-licensed. Note: the bare npm name `pi-dispatch` belongs to an unrelated package — the scoped `@edgehero/pi-dispatch` is the correct install.

Componentry

Beautiful animated UI components for React.

⚠ Rykuuun shared it as a front-end dev finding; details unverified beyond the site's tagline.

Mint Playground

The public gallery for Mint's 3D-asset-creation tooling for AI agents: use Codex, Claude, or Cursor (via the Mint MCP server and Mint Skills) to generate, rig, and animate 3D models, build 3D worlds, and create PBR materials and asset packs. The playground showcases dozens of agent-built interactive 3D scenes, from mini-games to product configurators and data visualizations — though it is primarily a showcase for Mint's tooling rather than a tool you operate directly.

delphitools

A handmade-web collection of small, single-purpose browser tools: colour utilities (contrast checker, palette generator/extractor, Tailwind shade scales), image tools (compressor, converter, background remover, favicon generator, SVG optimiser, metadata stripper), plus audio/video and typography tools. Everything is free with no logins, no paywalls, and no data collection; fully open source (0-BSD), with a native iOS app and an offline CLI offering the same tools.

modly

An open-source desktop app (Windows, Linux, Apple Silicon macOS) that turns a photo or text prompt into a 3D mesh using open-source AI models running entirely on your own GPU — no cloud inference. It ships an extension system for swapping in models like Hunyuan3D 2 and TripoSG, node-based generation workflows, and a CLI so agents and scripts can drive generation headlessly; MIT-licensed, around 7.7k stars.

img2threejs

A skill for coding agents (Claude Code, Codex, OpenCode) that rebuilds the object in a reference photo as a code-only, procedural Three.js model — reconstruction-by-code, not photogrammetry or mesh extraction — with a staged sculpting pipeline (blockout → material → lighting → optimization) and vision-review quality gates. It is deliberately token-efficient: deterministic scripts handle validation while model tokens are spent only on visual judgment and code; Apache 2.0, around 16.7k stars, with a live demo gallery.

pi-openai-server-compaction

Shared without description; the name suggests an OpenAI-compatible server with context compaction for the pi agent.

⚠ Unverified.

Fractal (Plasma AI)

Plasma AI's Fractal wraps Claude Code and Codex in bounded recursive loops.

⚠ Details beyond the truncated preview are unverified.

freerange

From chenglou (React core team alumnus): a static verifier that checks ordinary TypeScript layout code at compile time, without a layout engine. ~650 stars.

60fps.design

An endlessly browsable gallery of ~2,000+ UI/UX animation shots from best-in-class mobile and web apps — buttons, sheets, onboarding, loading states, micro-interactions — tagged across 108 filters. Useful swipe-reference for motion details before implementing your own.

pi-queue-steer

A pi extension adding a Cursor-style visible follow-up queue: pending agent tasks are listed, inline-editable, and steerable from a terminal UI while pi works. Small repo (~35 stars) but fills a real gap in agent-driven workflows.

pi-grok

Use your x.com Premium/SuperGrok subscription in pi with OAuth login, auto-refresh, and live model discovery; supports grok-4.5, grok-composer-2.5-fast, grok-build, grok-4.3, and x_search.

⚠ Feature claims come from the repo description and are unverified.

Kimi-Vendor-Verifier

A vendor verifier from MoonshotAI; baanish asked if anyone has run it.

⚠ Details unverified from the message text.

agentic_coding_flywheel_setup

An agentic coding setup that Rykuuun found "more flexible" than alternatives.

⚠ Details unverified from the message text.

tailscale-shell

A zsh-based proxied shell plus exit-node picker for Tailscale userspace networking: pick an exit node by country, city, or code and route egress through Mullvad or your own relay via SOCKS5. Small repo (~4 stars), but a practical wrapper around the tailscale CLI.

Dory

Docker & Linux containers native to Mac — free, open source, one shared VM for all containers, claiming ~4.7x less idle memory than per-container setups.

⚠ The memory claim is the vendor's and unverified.

archify

An agent skill (Claude/Codex/OpenCode compatible) that generates architecture, workflow, sequence, data-flow, and lifecycle diagrams as self-contained HTML with motion, then exports crisp PNG/JPEG. ~72k stars; positions itself as a verifiable, agentic alternative to Mermaid.

beads_rust

A fast Rust port of Steve Yegge's beads — a local-first, non-invasive issue tracker storing tasks in SQLite. Rykuuun "really loved this tool" and says it makes things easier to manage.

Loop Library (ForwardFuture Signals)

A library of 85 copy-pasteable AI agent "loops" from ForwardFuture — repeatable workflows like the docs sweep, the architecture-satisfaction refactor loop, the production-error sweep, and the SEO/GEO visibility loop — each with explicit checks and stopping conditions. Searchable across engineering, design, content, operations, and evaluation categories.

Flue 1.0 Beta

Flue 1.0 Beta announcement — "the open framework for building agents without lab lock-in: write once, deploy..." Details beyond the embed preview are unverified.

LMCache

"Supercharge Your LLM with the Fastest KV Cache Layer." The performance claim is the project's and unverified.

pi-shell-acp

A Herdr/tmux setup where sibling AI sessions — pi, Claude Code, Codex, Copilot, OMP, Antigravity — can message and spawn each other while keeping their own auth, tools, and transcripts. The repo has since been renamed from pi-shell-acp to "entwurf" and now covers the Agent Client Protocol and MCP agent protocols more broadly.

es-git PR #213

A fix by monotykamary — diff.print() (default Patch format) was stripping the +/-/space line-origin prefixes from output; this PR prepends them back.

pi-tool-repair

A validate-then-repair extension for pi that fixes common LLM tool-call mistakes (null fields, stringified arrays, wrong field names, anchor bleed) before tools execute. Tom: "technically their tool repair brought over to pi (+ my additions)."

fff PR #593

A fix for fff — the /fff-mode override was only stored in an in-memory variable lost on /reload and session restart; this PR persists fff-mode across reloads and session resumes.

fff

"The fastest and the most accurate file search toolkit for AI agents, Neovim, Rust, C, and NodeJS." The performance claim is the project's and unverified.

ccusage

Analyze coding (agent) CLI token usage and costs from local data.

hyperframes

HeyGen's TypeScript framework for generating video from HTML — describe motion with GSAP-style animations, render via Puppeteer/ffmpeg, and expose it to coding agents over MCP. The linked framemd section documents a Markdown-first animation format so agents can author motion graphics programmatically; ~53k stars.

MoneyPrinterTurbo

One-click HD short-video generator: feed it a topic or keyword and an automated AI workflow scripts the video, generates subtitles and TTS narration, and renders a finished clip for Reels/Shorts/TikTok. ~126k stars; note the repo docs are largely in Chinese.

Cozy

A local-first private journal: entries live as plain files in a folder on your own device — no cloud, no account, no AI reading your diary — with photos, audio, tags, mood/location tracking, and password-lockable journals. Imports from Notion, Day One, Obsidian, and Bear; $49 one-time ($89 forever) for macOS, Windows, iPhone, and iPad.

MiniMax M3

MiniMax's M3 announcement — frontier coding capability, 1M context, native multimodality, and a new MSA architecture.

⚠ All vendor claims, unverified.

codex-lb

A Codex/ChatGPT multiple-account load balancer and proxy with usage tracking, a dashboard, and OpenCode compatibility. Tom: "gonna try first."

codex-auth

A Zig-written CLI for switching and managing multiple Codex accounts — handy for anyone juggling several Codex subscriptions or auth contexts from the terminal. ~2.7k stars.

Zinley

"Built to extend you. 10× your day" — an AI extension of you across every device and person: takes your calls, writes your emails, and more.

⚠ Vendor claims, unverified.