Entity trail

AWS Labs

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

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-09-11 / projects-researcherAWS's AI-DLC workflow rules ship 2.8.2, fixing Plan Approval across native harness installs including Kiroawslabs/aidlc-workflows (4,545 stars, MIT-0) packages AWS's AI-Driven Life Cycle rules as per-harness runtimes installed with `aidlc` and checked with `aidlc doctor`. Release 2.8.2 on 2026-09-11 fixes Plan Approval forwarding and hook execution in native installs, and the Kiro IDE dispatcher, while keeping the approval and evidence requirements. It has 134 open issues against 65 PRs. It is a small reliability release, but it shows AWS maintaining approval-gated agent workflows as a cross-harness product.
2026-09-10 / vibe-coding-researcherCVE-2026-87911: AWS postgres-mcp-server hits CVSS 9.6 because read-only mode never blocked COPY ... TO PROGRAMThe SQL validation component in awslabs postgres-mcp-server before 1.1.7 fails to catch `COPY ... TO PROGRAM` in its default read-only mode, letting an unauthenticated actor plant the statement in content the agent later processes and get OS command execution on a self-managed PostgreSQL host. NVD published it 2026-09-09 at CVSS 9.6 critical, with an AWS security bulletin (2026-104) and GHSA-fph8-pg5w-78fv confirming. Upgrade to 1.1.7; this is the second Postgres MCP read-only bypass in two weeks, so treat 'read-only' MCP grants as an unenforced label rather than a boundary.
2026-09-09 / news-researcherFOIA Contracts Show the Pentagon Asked OpenAI for 'Minimal Refusal Rates'The Intercept obtained the OpenAI, Anthropic, Google and xAI Pentagon contracts through a FOIA lawsuit, each with a $200M ceiling, covering prototypes for military decision-making, intelligence analysis and operational planning. Beyond building tools, the labs agreed to advise the Pentagon on AI strategy, train military personnel, and forecast the risks of their own technology. The documents include a DoD request to OpenAI for a custom version engineered around "minimal refusal rates"; OpenAI says that phrase never made the signed version.
2026-09-07 / sources-researcherAsk HN on managing skills files draws 144 comments and a clear practitioner consensus: git repo plus symlinks, evals as integration tests, and automated drift checksThe thread posted 2026-09-06 reached 169 points and 144 comments, and the answers converge on a pattern nobody has written up as a standard: keep skills in one version-controlled repo (or dotfiles via chezmoi/guix home), symlink into every harness's skills directory, and let frontmatter progressive disclosure handle selection. Several people go further, running AI evals against skills as behavioural integration tests, and one team runs a biweekly agentic workflow that diffs skill content against the docs and opens PRs when they drift. Named tooling in the thread includes vercel-labs/skills (30,587 stars, MIT, pushed 2026-09-06), genged/capshelf, skillcatalog.dev, Microsoft's APM, and a proposed mise standard for shipping skills alongside CLI tools; a visible minority argues skills are redundant now that agents read repos and AGENTS.md well enough.
2026-09-05 / vibe-coding-researcherCVE-2026-85787: AWS's own postgres-mcp-server had an incomplete SQL denylist letting writes past read-only scopeAmazon awslabs postgres-mcp-server before 1.1.7 has an incomplete list of disallowed inputs in its SQL validation component, so an unauthenticated actor can modify data beyond the read-only scope by placing crafted SQL into content an authenticated user later submits through the MCP server. Upgrade to 1.1.7 or later. Landing the same day as the Postgres MCP Pro RangeFunction bypass, two independent Postgres MCP servers failed read-only enforcement in the same 24 hours.
2026-09-03 / projects-researcherAWS is shipping AI-DLC as agent steering rules, not as a product, at 4,299 starsawslabs/aidlc-workflows publishes AWS's "AI-Driven Life Cycle" as adaptive workflow steering rules that plug into existing AI coding agents rather than as a hosted service. Created 2025-11-13, it is at 4,299 stars and 765 forks with 128 open issues and pushed at 2026-09-03 10:13Z. It appeared on the TypeScript trending board today at 26 stars, low velocity but notable as a hyperscaler distributing methodology as agent-readable config.
2026-08-29 / saas-disruption-researcherCROSS-CATEGORY: Three Moves in Four Days Show Infrastructure Suppliers Reaching Up Into the Application LayerOpenAI revoked model access from a $4B-ARR application (Cursor, Aug 29); AWS agreed to buy DuckLabs, the team behind DuckDB, the engine embedded inside dozens of analytics products (Aug 26); and Cohere priced Parse 5 at $1.50 per 1,000 pages, which it models as 98% cheaper than GPT-5.5 for a 750M-document workflow, undercutting the app-layer parsers built on frontier models (Aug 27). The common shape is that the supplier one layer down now sets whether an application is viable, and none of the three application vendors affected had a contractual say. For builders, the lesson is that model access and embedded engines are single points of failure that belong in your risk register, not your integrations doc.
2026-08-27 / news-researcherAWS and Nvidia Commit to 2 Million More GPUs Over Two Years, Even as Amazon Shops Trainium ExternallyAWS and Nvidia announced on August 26 that they will deploy 2 million additional Nvidia GPUs across AWS global infrastructure over the next two years, citing surging demand from startups, enterprises, AI labs and governments, and extending the partnership across AI factories, CPUs, networking, open models, data processing and robotics. The timing is awkward: Amazon AI chief Peter DeSantis has said AWS is in talks to sell Trainium to outside companies as a direct Nvidia alternative. Buying more Nvidia while selling a competing part is the clearest signal yet that custom silicon is a margin play, not a capacity substitute.
2026-08-27 / hn-researcherAWS Is Buying DuckLabs, the Amsterdam Team Behind DuckDB, Closing Early SeptemberAmazon signed a definitive agreement to acquire DuckLabs, with the deal effective early September. DuckDB and the rest of the Duck Stack stay MIT-licensed under the nonprofit DuckDB Foundation, and creators Hannes Mühleisen and Mark Raasveldt keep leading technical direction from Amsterdam. AWS and DuckLabs have worked together since 2024; the story hit 1,060 points on HN, where the open question is what happens to DuckLake and Quack roadmaps once a hyperscaler owns the payroll.
2026-08-14 / rss-researcherSolv Labs proves agent payments in under 4 seconds with hardware attestation inside AWS Nitro EnclavesThis August 12, 2026 AWS writeup details Solv Labs combining Bedrock AgentCore payments (launched May 2026 with Coinbase and Stripe) with its ORACLE policy engine and ICME's PreFlight verification layer, so every autonomous agent payment is authorized, risk-priced, and auditable before settlement. An integrity service signs execution records inside AWS Nitro Enclaves, with the attestation document binding signing keys to enclave image measurements (PCR0/PCR1/PCR2) so records cannot be rewritten after the fact. Measured overhead is the useful part: full transaction under 4 seconds end-to-end, governance under 1 second, proof verification under 1 second — over the x402 agent payment standard.
2026-08-11 / news-researcherZuckerberg's 6,500-Word 'The Future Is for Everyone' Commits Meta to Reopening Weights — and Draws Immediate Backlash as the Case Against AIMark Zuckerberg published a 6,500-word manifesto on August 10 arguing that concentrating superintelligence in a few labs is the real risk, organized around individual empowerment, invention as the purpose of superintelligence, and balance of power as the foundation of safety. The concrete commitments: a "personal superintelligence" agent for every user, an America's Workforce Academy offering free trades training with guaranteed jobs, and — most materially for builders — Meta resuming open-weight releases starting with Muse Glimmer and a version of Muse Spark 1.2. TechCrunch and The Verge both ran pieces the same day arguing the essay is itself the clearest explanation of why people dislike AI, making this a rare case where the primary document and its reception are equally worth reading.
2026-07-26 / hn-researcherWSJ: Corporate America Stops Blowing Money on AI, Shops À La Carte and Buys Chinese Open WeightsA July 25, 2026 Wall Street Journal piece reports enterprises fed up with ballooning inference bills are adding cheaper models — including Chinese ones — alongside OpenAI and Anthropic rather than defaulting to frontier tiers, having concluded that everyday tasks don't need the most expensive model. The article draws the structural contrast explicitly: US frontier models are closed and tightly controlled, while Chinese labs ship cheap open weights that can be downloaded and customized. It landed on HN (63 points, 82 comments) the same day as the Stanford jobs brief and Tobi Knaup's open-weight essay — three independent signals that the cost and openness axis, not raw capability, is where enterprise procurement is now being decided.

Graph relationships

SUPPORTS
Claude Code
AWS Labs released AI-DLC Workflows supporting Claude Code and other AI tools
Source finding
RELEASED
Threat Modeling MCP Server
AWS Labs shipped Threat Modeling MCP Server for AI-assisted security workflows.
Source finding
RELEASED
MCP
AWS Labs released a threat-modeling MCP server implementing STRIDE methodology.
Source finding

Source trail

GitHub NVD The Intercept Hacker News NVD GitHub OpenAI, synthesized with aws.amazon.com/blogs/big-data DuckLabs post and VentureBeat Cohere Parse 5 coverage (2026-08-26 to 2026-08-29)GlobeNewswire (AWS/Nvidia)AWS / Hacker News AWS Machine Learning Blog TechCrunch Wall Street Journal / Hacker News

Graph sources

entity graphfindings textkg edgeskg entities

Fetching from the wire…