Dedoomify is a tool that rewrites AI news stories to describe failures like ordinary software issues, for example turning "the model is misaligned" into "the model has a bug." Each change is highlighted with the original text available on hover or tap, and it runs on quick phrase rules or a local AI model in the browser via WebGPU. It is open source on GitHub.
agent-blast-radius is an Agent Skill that reads an AI agent's codebase and reports two things: its boundary (whether anything needs or reaches the outside world, such as hosted model APIs, telemetry, start-time package or model fetches, SaaS connectors and hard-coded hosts) and its blast radius (what the agent can do inside an enclave if steered by a poisoned document, an insider or a compromised host). It writes a BLAST_RADIUS.md report with file:line evidence, matched failure patterns, OWASP agentic (ASI) ids and the smallest fix per path, backed by a deterministic, read-only, offline Python scanner that never runs the target code or prints secret values. It installs as a Claude Code plugin (2.1.275+) or manually into any Agent Skills host, and is static-only, so network policy, IAM policies, token scopes and runtime MCP tools go under "Unverifiable statically."
RefTrail is a method and toolset that makes a coding agent's completion report checkable: each requirement gets a source and measurable acceptance criterion, the criteria are frozen with a SHA-256 digest before product code is written, and a coverage catalog records status, evidence and the limits of that evidence. Its tools require Node.js 22.16 or newer and no further packages, and a worked example rebuild of GNU uniq reports 100% functional coverage with 1/1 mandatory checks passed, with recorded output comparisons matching GNU uniq byte for byte (16 of 16 under Linux with version 9.4 for revision 1; for revision 2, 16 of 16 under Windows with version 8.32 and 16 of 16 under Linux with version 9.4). The project states that it organizes what you are allowed to observe rather than analyzing products, that the digest proves criteria did not change but not that they preceded the code, and that its legal notes are general information, not legal advice.
Sploom is a tool that lives alongside git and records which agent and model wrote each line of code, along with the intent, prompts, decisions and checks behind it, working with Claude Code, Codex, Cursor, Gemini CLI, Copilot CLI, Devin CLI, Grok Build, Goose, Cline and any MCP-capable agent. After a one-time `sploom init`, it installs git hooks so provenance travels with commits via git notes, and offers commands like `sploom why`, `sploom rewind --green` and `sploom context`. Prompts are encrypted to the team's SSH keys listed in a committed `.sploomkeys` file, and it runs locally with an initial ~130 MB embedding model download.
OpenShorts is an MIT-licensed, self-hostable tool that turns long videos into 9:16 shorts using Gemini for viral moment detection, faster-whisper subtitles, and AI reframing modes including face tracking and split-screen layouts. It also bundles an AI Shorts UGC creator and a YouTube Studio toolkit, and exposes the pipeline via an MCP server, REST API, webhooks and CLI. Self-hosting is free with bring-your-own API keys, while a hosted cloud plan starts at $12/mo after a watermarked 20 min/mo free tier.
AWS introduces the Universal Agent Evaluation Framework (UAEF), an open-source, framework-agnostic system that uses adapters to normalize agent traces from LangGraph, LangChain, Strands Agents, Amazon Bedrock AgentCore, and Langfuse into a common AgentTrace format, then scores them with 36 built-in metrics across 7 quality dimensions. Results are stored per experiment in Amazon DynamoDB and Amazon S3, enabling comparison of agent versions, regression detection, and report and insight generation, with both a Python library and a low-code web UI sharing the same engine. The post also provides a playbook for choosing metrics and preparing datasets, plus runnable notebooks for each supported framework.
Why it matters: It offers a detailed, practical reference for teams evaluating agents across frameworks, including metric selection guidance and a closed-loop workflow for catching regressions before release.
AI Code Rating is a proposed standard in which a project places an ACR.md file in its repo root with a three-character rating: an uppercase letter for the maintainer's expertise (A–E), a digit for the share of code written by AI (0–4), and a lowercase letter for how closely a human checked AI-written code (a–e). The site offers a rating form that generates the badge and ACR.md file, a validator, and a GitHub Action, and it lists adopters including aicoderating.com, DomainMOD, Email Health Check, and HomelabAPI. The spec is at version 0.1, and the roadmap covers publishing the spec, tooling, adoption, and proposing that code hosts display the rating.
Tensward is an Apache-2.0 tool that measures how your LLM is served on your own machine by starting its own copy of vLLM or llama.cpp on a local port, sending a realistic workload, and reading engine and GPU metrics. It offers one flag change at a time, measures it, compares answers before and after, and keeps the change only if it is faster and the answers pass your check; it never stops or reconfigures your running server. It supports NVIDIA and Apple Silicon hardware, and the post includes two L4 case studies showing gains such as +40% requests/s and first word −47% on llama.cpp, and first word 19 s → 0.30 s on vLLM.
Hubzoid is an Apache-2.0 tool that turns a folder of Markdown files—knowledge, rules, skills and agent instructions—into a shareable AI agent usable in chat, repeatable workflows, or via MCP in Claude Code, ChatGPT or Codex. It runs on an existing Claude Code or Codex login (or a model API key), supports team sign-in, per-person and per-agent access, approvals before consequential actions, and a record of every run. The starter hub is a fictional shop whose refund rule in knowledge/house-rules.md can be edited to see answers change, with evals to check they stay correct.
Neo4j released Relate, an open source (Apache 2.0), in-memory, in-process embedded database built in Rust for AI agents and agentic systems, which installs quickly and starts in milliseconds. It offers built-in MCP server connectivity, a CLI, Bolt and HTTP Query API v2 endpoints, sub-millisecond in-memory reads, and is 100% openCypher and more than 97% GQL compliant. Relate is available at https://relatedb.ai.
Episteme AI is a Next.js app that turns career goals into time-bounded, prerequisite-ordered weekly roadmaps built from free public courseware, using Gemini 2.0 Flash with a deterministic 52-week fallback. It uses Cheerio to detect paywalls and dead links via Schema.org JSON-LD and HTTP checks, then swaps in verified free alternatives, with approval-gated calendar and audit-link dispatch backed by SQLite idempotency keys. A live demo is available at episteme-ai-one.vercel.app.
Editmamei is a desktop MCP server that lets Claude and other MCP-compatible AI apps drive real Photoshop 2026 or GIMP 3.2 tools layer by layer, keeping edits reversible and leaving final touches to the user. It ships 112 tools (83 for Photoshop, 29 for GIMP), runs on-device vision models in Photoshop, and offers a free Community edition plus a paid Pro module. A hosted Editmamei Connect version edits photos on its servers, with uploads deleted after 24 hours.
Kodelyth ECC v2.31.1 is an MIT-licensed, locally run toolkit offering 71 specialist AI agents, 196 skills, 104 slash commands, and an MCP server that auto-wires into six IDEs including Claude Code, Cursor, and Windsurf. It claims token savings from RTK input compression (60–90% fewer input tokens), Terse output compression (40–70%), and a tree-sitter codebase graph across 158 languages, with zero telemetry. It installs via `npm i -g kodelyth-ecc` and also includes a Super Agent orchestrator, Devil Mode red-team crew, and local BM25 Memory Core.
Model Desk is an open-source, MIT-licensed Linux desktop app built with Rust and Tauri that connects to Lumo via lumo-tamer, DeepSeek, and any OpenAI-compatible endpoint, with an OpenCode coding agent, local history, attachments with local OCR, tools with permissions, memory, and a Bonsai image studio. Version 0.4.1 adds automatic Bonsai backend start, Lumo Tamer, DeepSeek, and the OpenCode agent, and is distributed as an AppImage or .deb for Linux x86_64. The app stores data in ~/.config/model-desk and reports no telemetry.
Agent Buddies is a library of ten original animated characters that each express the same eight agent states, driven by a deterministic event stream with no API calls. It ships React, Web Components and SwiftUI support under MIT code with openly licensed art, plus a picker that lets end users browse, preview and save a favorite buddy on their device. Artists can create their own characters by defining regions on a master image and validating a pack, with a CLI preview available via npx while the npm release awaits owner review.
Lokey is a native Kotlin Android keyboard that runs word prediction, autocorrection, dictation and translation models locally on the phone, without an account, subscription or inference service connection. It supports 39 languages and eleven keyboard families, with optional models including LFM 2.5 350M, Qwen 3.5 0.8B, Parakeet v3 and Qwen3 ASR, downloaded separately from the app. It requires Android 9 or later on an ARM64 device, is beta software, and is released under the MIT license with third-party runtimes and model weights under their own licenses.
vLLM details its Encoder/Prefill/Decode (EPD) disaggregation work, including metadata-only requests that let downstream instances reuse remote embeddings without repeating media preprocessing, a dedicated encoder-only Model Runner V2 path, and remote embedding transfer via NIXL and Mooncake. It also covers cross-encoder output reuse through Mooncake Store and orchestration support in vLLM Router, SMG, and llm-d, with core capabilities in vLLM v0.32.0 and DeepSeek V4.1 Flash metadata-only support on main via #62802. A DeepSeek V4.1 Flash benchmark on GB200 found EPD delivered 6.6–9.5% higher per-GPU throughput and 44–47% lower P90 TTFT than PD at concurrency 64–128, but lower per-GPU throughput at concurrency 16.
MBZUAI's Institute of Foundation Models released PAN 2, an Apache 2.0 world model that takes a scene and text action and simulates the outcome as video, shipping weights, training code and inference code. It pairs an LLM planner with a 29B MMDiT diffusion transformer trained from scratch, outputting 704×1248 video at 24 fps up to 20 seconds. IFM reports 34.0% full-plan completion on 100 multi-step tasks, leading six tested models, while trailing MiniMax-H3 on single actions and human preference.
Ant Group's Robbyant has open-sourced a tactile AI model for robot hands, according to TechNode. The article is available at the linked URL.
The author presents Restless, a system for running a small "company" of AI agents inside a persistent Linux machine, where human users set direction and agents handle routine work such as research, builds, QA and scheduling. The post walks through an illustrative day at a fictional game studio, describing roles for people and agents, approval gates for external actions, model choice per agent, and pricing tiers. It also cites FrankenSurf, an open-source project with 100 GitHub stars, as a real example of a Restless company writing its code, shipping releases and marketing it.
ExecuTorch's new WebGPU backend runs PyTorch models exported with torch.export to a .pte file directly on the GPU in WebGPU-enabled browsers, without converting to another framework's format. The same artifact runs across Chrome and Safari on Metal, Vulkan and D3D12, and on an Apple M1 Pro in Chrome it was 4–7× faster than ExecuTorch's Wasm (XNNPACK) CPU path on prefill and 2–10× faster on decode. Validated models include Llama 3.2, Qwen3, Gemma 3, SmolVLM, Depth Anything V2, SAM 2, Whisper and on-device LoRA fine-tuning, all running locally in the tab.
Why it matters: It documents a concrete path for running PyTorch models on browser GPUs with benchmark numbers and a list of validated models, useful for developers weighing in-browser inference.
PushForm is an open-source (GNU GPLv3) 3D modeler that opens and exports SketchUp .skp files and LayOut documents, and imports/exports formats including STEP, glTF, COLLADA, OBJ, 3DS, STL and DXF. It offers push/pull modeling with exact dimensions, multiple display styles, and automatic cloud saving, aimed at cabinet makers, woodworkers, interior designers and others. The source is available on GitHub.
AgentGuard is an open-source (Apache 2.0) transaction guardrail layer for AI agents holding crypto wallets: before signing or broadcasting, an agent submits the proposed transaction to AgentGuard, which simulates it against live chain state, evaluates the effects against an operator-defined policy, and returns allow, block or escalate with an explanation, a 0–100 risk score and an immutable audit record. It supports EVM chains and Solana, decodes common drain vectors (unlimited approvals, setApprovalForAll, permits, zero-minimum-output swaps, SetAuthority, never-seen contracts), and ships a conservative default policy plus a TypeScript SDK that wraps any signer so signTransaction/sendTransaction are checked transparently. An optional separately licensed @agentguard/behavior package adds per-agent behavioral anomaly scoring, and both it and the reputation layer fail closed on outages or unreadable profiles.
Notebook AI is a desktop-first AI note taker built as a monorepo with an Electron/React/Vite desktop app, a Next.js web app, a local FastAPI/LangChain sidecar, and a hosted FastAPI cloud API, released under the MIT License. Notes are stored as markdown on disk via the local sidecar, and desktop notes, local provider keys, transcription, and AI work without signing in, while Google Drive mirror and the web account vault require an account. The project lists local and cloud API endpoints, setup guides, and macOS release build commands.
CodeLens is a code review tool that turns a pull request or local branch into a guided review: a brief of what changed and why, a queue of changes ordered by risk, and per-change context including callers, entry points, and external effects. It parses 15 languages with Tree-sitter (with TypeScript compiler type info for TS/JS), supports Claude, any OpenAI-compatible API, or local Ollama models, and caches results by content hash in SQLite or Postgres. It also offers codebase exploration, GitHub PR review with inline comments, a GitHub Action, and a heuristic offline mode; it requires Node.js 22.13+ and git and is Apache-2.0 licensed.
A blog post argues that when an AI agent implements, tests, reviews and reports on its own code, its claim of "verified" is untrustworthy, and proposes binding evidence (review, tests, human approval) to a specific Git tree hash so that changing the tree invalidates the evidence. It introduces Uncle (github.com/unclehq/uncle), an open-source tool that runs checks itself, records exit codes and tree identities, and distinguishes "verified," "signed" and "human looked at it" rather than blurring them. The author notes the tool does not establish correctness, is young, and is partly defeated by a hostile agent with unsandboxed shell access.
aisync is a local hub (~/.ai-sync/, a git repo) that lets Claude Code, Codex and Cursor share rules, a hash-chained journal and file claims, with an overlapping claim exiting with code 3 so an agent stops before colliding. Hooks inject what other tools just did into running sessions, and version 0.2.0 adds layered When/Do/Not rules, a progress_file for stable prompt caching, and an optional MCP server. It requires Python 3.11+ and git on macOS or Linux, uses only the standard library, and is MIT licensed.
Latent Space benchmarks NaiveAI's open-weights Naive-N0.5-Flash, a MiMo-V2.5 derivative adapted to DeepSeek Sparse Attention with about 3.25T extra training tokens, against DeepSeek V4.1 Flash 1017 and GLM 5.3 Flash on speed, verbosity and three app-building prompts. It reports peak throughput near 2,100 tokens/s and a median response time of 3.1 seconds versus 5.3 and 10.1 seconds, but far higher thinking-token use on hard questions and mean times similar to DeepSeek. In the apps, Naive-N0.5-Flash produced the best solar system, tied DeepSeek on the tree model and trailed it on ray-tracer efficiency, while its no-thinking variant failed to build a usable tree app.
Woolly is a local, private memory layer that runs on your computer and gives MCP apps like Claude Code, Claude Desktop and Cursor one shared memory stored as plain Markdown files. It also adds memory to Ollama and OpenAI/Anthropic-compatible apps via `woolly proxy`, keeps facts current by replacing outdated ones, and lets you review, edit or delete memories through `woolly ui`. Installation is via a curl script, Homebrew, Windows PowerShell, or `go install`, with a Python SDK available as `pip install woolly-ai`.
Qitt is a proactive chief of staff built from T3 Code threads and scheduled tasks that watches incoming requests, triages them, and starts research worker threads ahead of time, but never sends anything—every output is proposed text for the owner to review. It is vendor-agnostic, customizable via Markdown charters and a local context file, and extendable through input-source folders, with Slack and JIRA shipping as examples; it has run on Claude only so far. It requires a T3 Code server and Node.js 22.19 or later, and runs on one always-on machine, losing events that arrive while it is down by design.
Relay is a six-role multi-agent pipeline (PM, Architect, Implementer, Reviewer, Prod-Readiness, Verifier) for shipping frontend changes with AI assistance, with a human approval gate after the Architect and a Verifier that defaults to FAIL. It is tool-neutral via role contracts in .agents/, claims >90% cache hit on warm sessions, and offers a lighter /pipeline-lite mode that skips PM, Architect and Verifier but refuses tasks mentioning auth, payments, PII, tokens or cookies. It ships as a Claude Code plugin with slash commands, a GitHub Copilot paste workflow, and an AGENTS.md route, tuned for React 18, TypeScript 5, Vite, CSS Modules, Axios, Jest + RTL and Playwright.
Fix Gate is a free GitHub Action that verifies security-fix pull requests, including output from Copilot Autofix, Snyk Agent Fix or a person, by running Railo's rules over the base and head versions of changed files and posting five checks as a check run. It reports whether the targeted finding is gone, whether new findings appeared in the changed lines, whether changed files still parse, whether tests newly failed, and whether an added validator rejects a fixed list of known bypass payloads. The engine is not open source, runs only on pull_request events, and the guard row executes the pull request's validator function in a separate unprivileged subprocess rather than a container boundary.
Brigade is an MIT-licensed open-source product that gives you a team of AI teammates running the real Claude Code or Codex CLIs signed in with your own Claude or ChatGPT subscription accounts, each with a job, its own computer, Apps and memory. Decisions needing you land in one list on your phone, and it is free to self-host or $99 a month per workspace for Brigade Cloud. The repository includes the dashboard, API, runner, desktop app and deployment docs, and most commits come from Brigade's own teammates with a person approving every change.
Why it matters: Useful for readers weighing agent orchestration tools, since it details self-hosting, subscription-based usage and approval-gated writes.
The post describes ExecuTorch's new WebGPU backend, which lets PyTorch models exported once to a .pte file via torch.export run in any WebGPU-enabled browser through a WebAssembly runtime, covering LLMs, vision-language, segmentation, depth, detection, super-resolution, speech-to-text and on-device LoRA/DiReFT fine-tuning. On an Apple M1 Pro in Chrome, WebGPU was 4–7× faster than ExecuTorch's Wasm/XNNPACK CPU path on prefill and 2–10× faster on decode, with all data staying on the device. The backend is still in early development, and the post links an ExecuTorch WebGPU Playground and Hugging Face repos with .pte files and export recipes.
Featured coverage of this story: 《ExecuTorch WebGPU: PyTorch Models on the GPU, in the Browser》
Zyphra has released ZONOS3-Mini, an Apache 2.0 text-to-speech model that is a sparse Mixture of Experts with 800M total and about 100M active parameters, supporting zero-shot voice cloning and word-level delivery control. It runs via WebGPU in the browser, MLX on Apple Silicon, ggml on CPU/CUDA/Metal/Vulkan, and SGLang-Omni on NVIDIA GPUs, with Zyphra reporting 80 ms median time to first audio on one H100 at 8 concurrent requests. Zyphra reports it ranks first among open models on its self-run ZTTS2-Eval Capability Index at 66.2 in speaker-embedding mode, while reference-clip mode scores only 44.7.
Mitra-v2 is a second-generation tabular foundation model for classification and regression, pretrained entirely on synthetic tables with 10x as many rows and roughly 3x as many columns as Mitra-v1, plus a new hybrid SCM prior and hybrid optimizer. On TabArena it leads the single-model field, beating Mitra-v1 on 32 of 33 datasets where both run natively and completing all 51 datasets within the one-hour budget, and fine-tuned Mitra-v2 ranks first on ScoringBench. The fine-tuned models and inference/fine-tuning code are released under Apache-2.0 on Hugging Face.
Why it matters: Useful for readers tracking tabular foundation models, as it reports concrete benchmark rankings and an open Apache-2.0 release.
Dryplate MCP is an MCP server that gives agents in Claude, Cursor and other MCP clients local image tools: convert, compress, resize, crop, rotate, blur, watermark, merge and strip metadata, including reading iPhone HEIC, with nothing uploaded. It is part of a set of packages alongside @dryplate/sdk, a JavaScript SDK for Node (on sharp) and the browser (WASM codecs), with shared operations in packages/core. The project is MIT licensed, and the SDK ships libheif as an unmodified WASM file under LGPL-3.0.
Deedglass is an open-source, self-hosted tool that sits between AI agents and the systems they modify: rules decide each tool call, a check compares what the target actually changed against what the agent reported (Claimed, Checked or Proven), and each result is sealed as a signed record verifiable offline. Its demo shows an agent reporting a 1-row UPDATE while a legacy trigger changed 301 rows in the same transaction, caught via Postgres logical decoding, with an undo that runs only after a person countersigns. It is pre-1.0, has one maintainer, and Proven checks cover Postgres only, requiring wal_level=logical and REPLICA IDENTITY FULL on watched tables.
Worktrunk is a CLI for Git worktree management designed for running parallel AI agent workflows, offering three core commands (switch, list, remove) that address worktrees by branch name. It adds hooks for workflow automation, LLM-generated commit messages, a one-command merge workflow, an interactive picker with CI status, and copy-on-write build cache sharing on APFS, btrfs, and XFS. It is installable via Homebrew, Cargo, Winget, Arch Linux, and Conda/Pixi.
NVIDIA Snapshot is a new open-source component of the NVIDIA DSX AI Factory Platform that makes initialized GPU workloads reusable across inference replicas, speeding startup and recovery. The post notes that when inference demand spikes or a replica fails, available GPUs do not immediately translate into serving capacity, since initializing an LLM workload—a cold start—can take minutes.