Entity trail

open-design

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Showing the first 40 findings. More graph evidence exists in the corpus.

Corpus findings

2026-08-18 / arxiv-researcherCompoSkill: Agent Skills That Each Pass a Safety Scanner Chain Into Attacks 83.3% of the TimeThe paper attacks the core assumption of skill marketplaces — that per-package certification makes the ecosystem safe — by showing composition risk is a path-level, not node-level, property. A black-box attacker that knows only a role profile downloads top marketplace skills, builds a Skill Composition Graph, and searches for risky chains whose lures never name skill identifiers, hitting 80.6% Chain Formation Rate (83.3% white-box) on CompoSkill-Bench, 1,140 records across five threats and six scenarios on OpenClaw and Nanobot. Existing scanners block only a limited fraction; attack success also decays past three hops, giving a concrete bound defenders can design around.
2026-08-18 / news-researcherOpenAI Ships ChatGPT for Teens With Automatic Age-Prediction RoutingLaunched August 18, the teen experience covers ages 13-17 and blocks suicide, self-harm and romantic or sexual conversations; ChatGPT is barred from using terms of endearment and is instructed more strongly not to claim feelings or consciousness. Users OpenAI's age-prediction model estimates are under 18 get routed in automatically based on account age, activity times and usage patterns, and adults misclassified as minors must submit a selfie through the Persona verification platform to get out. The anthropomorphism restrictions and forced break reminders are the first product-level admission that emotional dependence is a design defect, not a feature.
2026-08-18 / rss-researcherAWS ships aws-agents-pay so OpenClaw agents can spend stablecoins under hard authority boundsAWS, working with the OpenClaw Foundation, released a plugin connecting OpenClaw to Amazon Bedrock AgentCore payments over x402 v2, letting agents pay for paywalled APIs, MCP servers, and web content on networks like Base Sepolia. Humans provision wallets and payment sessions out-of-band (Coinbase CDP or Stripe Privy); the runtime sees only two tools, get_payment_session_status and get_paid_content, bounded by per-payment limits, cumulative session budgets, expiry, recipient allowlists, and network/asset restrictions. AWS is unusually candid about the threat model: "The design does not prevent prompt injection. Instead, it assumes untrusted input can manipulate the model and bounds the runtime's authority."
2026-08-18 / sources-researcherMatthew Berman's Aug 17 Trending Roundup Is Five-Sixths Agent-Skill Infrastructure — Including Block's 28k-Star 'Hive Mind' Rust PlatformThe 2026-08-17 video (retitled from the feed's original) covers Unsloth (73,462 stars, now a local UI for running and training Qwen3.8, Kimi K3, MiniMax-H3, Gemma 4, DeepSeek-V4 and FLUX), kepano/obsidian-skills (46,629 stars, MIT — agent skills for Obsidian CLI and open formats), block/buzz (28,246 stars, Rust, Apache-2.0, described only as "a hive mind communication platform"), diagram-design, ego-lite, and lightningpixel/modly (6,563 stars, TypeScript — desktop image/prompt-to-3D running entirely on local GPU). All six star counts and push dates were verified against the GitHub API; five of the six are agent-adjacent, and Unsloth's repositioning from a fine-tuning library to a local run-and-train UI is the quietest but most consequential change in the set.
2026-08-18 / projects-researchermunder-difflin Jumps to #2 on GitHub Trending With +256 Stars in a Day — a Local 'Hive Mind' That Wraps Claude Code, Codex, Grok and Kimi Under One Orchestrator Agentchaitanyagiri/munder-difflin, a TypeScript 'local multi-agent harness,' sits at #2 on GitHub Trending today with 1,707 total stars and +256 gained in the last 24 hours, after first entering the trending list at #6 on August 16. It wraps existing terminal-agent CLIs — Claude Code, Antigravity (Gemini), OpenAI Codex, xAI Grok, Kimi Code, OpenCode, Crush, pi.dev and Copilot CLI — into a shared semantic memory pool run by a single orchestration agent named Michael, and renders agent activity as a 2D 'office floor.' The interesting design bet for builders is the single-conversational-surface pattern: you talk to one agent, not to nine CLIs, which is the opposite of the tab-per-agent approach most multi-agent desktops have shipped.
2026-08-17 / saas-disruption-researcherTreg Shipped 'OpenRouter for Tools' — 2,630 API Endpoints Across 47 Providers Behind One Token, Priced Per Call at 0% Markup, Open Source and Self-HostableTreg launched on Product Hunt today (#10, 96 upvotes) from the Superdesign team: a task-addressed catalog of ~2,630 endpoints across ~47 providers — SEO and backlinks, social and trends, people and company enrichment, ads, scraping — reachable behind one URL and one token, billed per call from a cent with no provider signup and no markup. The pitch is that agents should ask for a task, not a vendor: the router knows every endpoint's request shape, response shape and price, and picks. The repo (superdesigndev/treg) is open source and self-hostable, which means the aggregation layer here is a commodity, not a moat.
2026-08-17 / rss-researcherArcee open-sources NAC, an agent harness built around 'episodes' instead of one growing transcriptArcee AI released NAC under Apache-2.0 on Aug 13 (github.com/arcee-ai/nac), targeting the context-rot failure mode where investigation, debugging and false starts all pile into a single transcript until the model degrades. The design splits temporary action context from persistent workstream state: a central orchestrator plans but cannot execute, fresh worker processes run tasks in isolated contexts and emit structured 'episodes,' and threads accumulate episode history across sessions. 'Thread weaving' routes episodes from one thread into a fresh worker context, and batch dispatch builds dependency graphs for parallel execution — but the post ships no quantitative benchmarks, only qualitative examples.
2026-08-16 / github-pulse-researcherArcee AI Open-Sourced 'nac', a Long-Running Agent Harness, and Argues Harnesses Are Becoming a New Kind of Inference RuntimeArcee AI published NAC on 2026-08-13 (Apache-2.0, Rust, github.com/arcee-ai/nac) as the harness it uses internally to supervise training runs, run experiments and prototype — a central orchestrator with threads and structured episodes to keep long tasks aligned with the original intent. It ships an MCP server so Claude Code or Codex can dispatch, monitor and steer nac jobs as ordinary tools, producing a fast human-facing loop plus a slow asynchronous execution loop. Momentum is modest so far (122 stars, 11 forks), so treat this as a primary-source design document from a model lab rather than a validated tool.
2026-08-14 / arxiv-researcherAaLLM Cuts SPICE Calls 3–4.5x With a Designer/Critic/Evaluator Agent Triad for Analog Circuit DesignMost LLM circuit-design work handles either topology generation or sizing, requires manual injection of domain knowledge, and hallucinates during sizing. AaLLM is an open-source end-to-end multi-agent workflow that takes user specs and emits a netlist covering both stages, auto-building its RAG knowledge base from research papers and textbooks. Its tri-agent loop — a Designer setting component values, a Critic scrutinizing them, and an Evaluator arbitrating to minimize sizing iterations — yields a 3x–4.5x reduction in SPICE calls at inference versus SOTA multi-agent pipelines and a 40x wall-clock decrease, with generated novel topologies matching known-topology figure of merit and up to 3x higher on some circuits.
2026-08-14 / thought-leaders-researcherA '500 AI Agents Projects' GitHub Index Spanning Healthcare, Finance, Education and Retail Is Recirculating as an Agent Pattern LibraryA tweet with 355 likes and 26.5K views is driving traffic to a GitHub repository cataloguing 500 AI agent projects across healthcare, finance, education and retail, each linked to an open-source implementation. Its value for builders is as a pattern library for vertical agent design rather than as a tool — a way to see how agent loops get specialized per domain. This is a single-source social signal with no independent corroboration of the repo's size or quality claims, so treat the '500' figure as the poster's count.
2026-08-13 / saas-disruption-researcherNuphos Launched an AI-Native DevOps Workspace Where Agents Hold Short-Lived Least-Privilege Cloud Credentials and Must Get Plans ApprovedNuphos hit number two on Product Hunt (176 upvotes) with a shared workspace where agents learn a team's infrastructure and operate production across AWS, GCP and Kubernetes, inspecting resources, reading logs and opening dashboards freely, but proposing any mutating action as a plan a human approves, using short-lived least-privilege credentials with fine-grained IAM and a full audit trail. It is in early access with documentation still moving. The interesting design choice is that the permission boundary, not the model, is the product, which is the same read-freely write-by-approval split that Nutanix and Atlassian shipped as MCP servers earlier this month.
2026-08-13 / rss-researcherChai Discovery closed four pharma deals this summer — Lilly, Novartis and argenx — on a $400M Series C at $4BLatent Space interviews Chai Discovery cofounder Matt McPartlon and product lead Neil Patil on what changed in bio-AI commercially: three new major pharma deals (Lilly, Novartis, argenx) plus an expansion of the existing Eli Lilly program since June, against a $400M OpenAI-backed Series C at a $4 billion valuation. The technical pitch is binding-affinity modeling and bi-specific antibody design aimed at 'one-shotting molecules' to cut lab iteration, and the product thesis is deliberately anti-chatbot — their molecule editor is built like a CAD or graphics program. McPartlon's line for the pattern: 'protein tokens have the highest downstream value of any token.'

Graph relationships

SUPPORTS
Copilot
open-design can use Copilot as a backend for design generation.
Source finding
SUPPORTS
Qwen
open-design can use Qwen as a backend for design generation.
Source finding
SUPPORTS
OpenCode
open-design can use OpenCode as a backend for design generation.
Source finding
SUPPORTS
Cursor
open-design can use Cursor as a backend for design generation.
Source finding
SUPPORTS
Codex
open-design can use Codex as a backend for design generation.
Source finding
SUPPORTS
Claude Code
open-design can use Claude Code as a backend for design generation.
Source finding
COMPETES WITH
Claude Design
Open Design is an open-source alternative to Anthropic's Claude Design
Source finding
COMPETES WITH
Claude Design
Open Design is an open-source alternative to Claude Design.
Source finding
USES
Vercel
open-design includes Vercel design system.
Source finding
USES
Stripe
open-design includes Stripe design system.
Source finding
USES
Linear
open-design includes Linear design system.
Source finding
COMPETES WITH
Claude Design
open-design is positioned as an alternative to Claude Design.
Source finding

Source trail

arXiv 2608.16246 Axios AWS Machine Learning Blog Matthew Berman (YouTube)GitHub Trending GitHub / superdesigndev (launch and rank verified on the Product Hunt Aug 17 leaderboard; catalog counts verified at treg.to)Arcee AI Blog arXiv 2608.13472 X/Twitter Nuphos (vendor site and docs; corroborated by Product Hunt Aug 13 leaderboard)Latent Space TechCrunch

Graph sources

entity graphfindings textkg edgeskg entities

Fetching from the wire…