Entity trail

Amazon Bedrock

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

2026-08-22 / rss-researcherAWS open-sources ADOP, a five-agent Bedrock reference architecture for Bronze-to-Gold data pipelinesThe Agentic Data Operations Platform, published August 2026, uses a Data Onboarding Agent orchestrator that spawns five specialists: Metadata (schema generation), Ontology (relationship inference), Quality (validation rules), ETL (transformation logic), and Orchestration (Airflow DAGs or Step Functions). It runs Claude Code through Amazon Bedrock using the Dynamic Workflow feature for sub-agent spawning. The framing is 'each new source is a prompt, not a project,' and the full reference implementation is on GitHub at aws-samples/sample-Agentic-Ai-Data-Operations. AWS gives no quantified cycle-time numbers, only a claim of significant reduction by Phase 3.
2026-08-21 / hn-researcherCodex CLI on AWS Bedrock Can't Opt Into Prompt Caching, and One User's Logs Show 85% of a $1,386 Bill Was Cache WritesGitHub issue openai/codex#37674 documents that Codex CLI 0.147.0 running GPT-5.6-Sol on Amazon Bedrock cannot pass 'prompt_cache_options' or 'prompt_cache_breakpoint', so every request performs a cache write instead of a cache read. The reporter's August 5-8 production logs show 3,656 requests generating 171.94M cache-write tokens, an estimated $1,182.09 of a $1,386.46 total, with local tests averaging roughly 88K wasted cache-write tokens per request. The thread hit 139 points and 54 comments and the issue is still open, assigned to celia-oai. Anyone running Codex through Bedrock should check their token telemetry before the next invoice.
2026-08-20 / rss-researcherGrok 4.6 hits GA on Amazon Bedrock with a 500k context window at $2/$6 per million tokensxAI announced on August 19 that Grok 4.6 is generally available on Amazon Bedrock across supported AWS regions, priced at $2 per million input tokens and $6 per million output. It carries a 500,000-token context window and four configurable reasoning levels (low, medium, high, xhigh), positioned for long-running agents and visual work. The pricing undercuts frontier Claude and GPT tiers on Bedrock, which matters if you are routing agent traffic by cost.
2026-08-20 / vibe-coding-researcherCodex 0.148.0 Ships Stable With `codex exec fork`, `/export` to Markdown and Bedrock as a Built-In ProviderOpenAI cut Codex rust-v0.148.0 stable on 2026-08-18 with roughly 245 commits, adding session forking via `codex exec fork`, archive/restore from the TUI resume picker, `/export` of a full TUI conversation to Markdown or clipboard, and Amazon Bedrock Runtime as a built-in provider with AWS profile and region support. Hooks can now run commands asynchronously and invoke MCP tools, and `/status` shows estimated thread credits or cost for eligible workspaces. Sandbox restrictions now fail closed on denied or unreadable paths on Linux and Windows, and the 0.149.0 alpha train opened the next day.
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."
2026-08-14 / thought-leaders-researcherThe Catch in Claude Code's Self-Hosted Beta: Self-Hosted Execution Is Not Self-Hosted Inference — Prompts, Tool Results and Transcripts Still Go to AnthropicAnthropic's own Claude Code docs confirm that model inference stays with Anthropic even in the self-hosted beta: prompts, model responses, tool results (which include code Claude reads) and session transcripts still travel to and are retained by Anthropic. Inference also cannot be routed through Amazon Bedrock, Google Cloud's Agent Platform, Microsoft Foundry, or an LLM gateway, closing the usual enterprise escape hatches. Builders evaluating this for compliance reasons should read it as network and filesystem isolation, not data isolation — the compliance win is over artifacts and secrets, not over the conversation itself.
2026-08-12 / rss-researcherPixieset's Bedrock alt-text feature holds 35% adoption 16 months on — one API call, 750,000 photos in week oneAWS published the case study 2026-08-11: Pixieset shipped an AI-generated alt text feature to millions of photographers in four months using Claude 3.5 Sonnet on Amazon Bedrock, generating alt text for over 750,000 photos in the first week and still seeing 35% of applicable users active 16 months after launch. Implementation was a single Bedrock API call dropped into an existing event-driven pipeline, with zero downtime since launch via cross-region inference and automatic failover to secondary models. The stated lesson is targeting selection, not modeling: they automated a genuine drudgery task rather than the creative work, which is why an audience the post calls 'among the most skeptical' for generative AI adopted it at all.
2026-08-11 / news-researchernOps Cut Agent Time-to-Production From 10-12 Months to Weeks by Abandoning Self-Managed LangChain/LangGraph on EKS for Bedrock AgentCoreAWS published a case study where FinOps company nOps rebuilt its Clara agent on Amazon Bedrock AgentCore, replacing a self-managed EKS stack running LangChain and LangGraph, and cut time-to-production by 75% from a 10-12 month baseline. The interesting part for builders isn't the AWS marketing number but what it implies about the orchestration-framework layer: the operational cost of running your own agent runtime (session state, memory, identity, scaling) is now large enough that teams are trading framework flexibility for a managed runtime. Treat the 75% as directional and vendor-sourced, but the migration direction is the signal.
2026-08-09 / agents-researcherBedrock AgentCore runtime instances go GA, letting long-running agents use full EC2 instance types instead of microVMsAWS announced on August 6 that runtime instances in Amazon Bedrock AgentCore are generally available, allowing agents to run on your own EC2 instances while AgentCore handles provisioning, patching, scaling and lifecycle management. It complements the existing microVM-based runtime option and targets sustained, resource-intensive or specialized-hardware agent workloads that need the breadth of EC2 instance types. This lands the same week AWS renamed the legacy service to Bedrock Agents Classic and closed it to new customers, making AgentCore the clear forward path for production deployments.
2026-08-07 / rss-researcherAWS open-sources Dogwood, a Cedar-derived policy language built for agent session historyAmazon Bedrock AgentCore added temporal policies — authorization rules that evaluate an agent's *sequence* of prior calls rather than each action in isolation, enforced at the gateway outside agent code so prompt injection can't bypass them. Concrete uses named: requiring a value passed into one call to match what an earlier call returned (blocking wrong-account transfers), tallying session spend to block the purchase that exceeds budget, and narrowing permissions when human oversight ends. The language behind it, Dogwood, is Apache 2.0 at github.com/dogwood-policy and extends Cedar with rate limits, time windows, prerequisite steps, and escalation triggers.
2026-08-06 / news-researcherAWS Publishes an MCP Bridge Letting Cloud-Hosted AgentCore Agents Call MCP Servers on a User's LaptopAWS detailed how it built a secure bridge so agents running on Amazon Bedrock AgentCore in the cloud can invoke MCP servers running locally on a user's machine, closing the gap where the agent is remote but the tools and files are local. The pattern matters because most enterprise MCP deployments today force a choice between local tools with a local agent or cloud tools with a cloud agent. AWS also published Mobileye's AgentCore support-operations deployment and a browser-based web insight extraction reference the same week, indicating a coordinated push behind AgentCore as the enterprise agent runtime.
2026-08-05 / rss-researcherAmazon Bedrock's Automated Reasoning policies now propose their own formal-logic fixesAWS added automatic policy refinement to Bedrock Guardrails' Automated Reasoning checks: when a test fails, a refinement engine diagnoses it and proposes formal-logic changes for human approval rather than requiring hand-edited rules. Two modes exist — iterative refinement for rule issues (translation succeeds but validation returns unexpected VALID/INVALID/SATISFIABLE/IMPOSSIBLE findings), and ambiguous variable refinement for TRANSLATION_AMBIGUOUS findings caused by overlapping or vague variable descriptions. The async workflow is export → start refinement → poll → retrieve proposed changes → review with test impact → accept or discard; every change requires explicit approval.

Graph relationships

BUILT BY
AWS
Amazon Bedrock is an AWS service
Source finding
USES
Capella iQ
Capella iQ runs on Amazon Bedrock
Source finding
SUPPORTS
OpenAI
Amazon Bedrock's reinforcement fine-tuning uses OpenAI-compatible APIs.
Source finding
USES
DevOps Agent
DevOps Agent integrates with Amazon Bedrock for agent management.
Source finding
SUPPORTS
Claude Opus 4.7
Claude Opus 4.7 available on Amazon Bedrock
Source finding
BUILT BY
AWS
AWS provides Amazon Bedrock
Source finding
USES
AWS
AWS architecture uses Amazon Bedrock for AI agents with persistent memory.
Source finding
SUPPORTS
MCP
Amazon Bedrock supports Model Context Protocol for agent tool use.
Source finding
RELEASED
Amazon
Amazon released serverless MCP proxy on Bedrock AgentCore Runtime.
Source finding
USES
Claude Apps Gateway
Claude Apps Gateway is deployable on Amazon Bedrock
Source finding
USES
HippoRAG
HippoRAG combines Bedrock LLMs for retrieval.
Source finding
USES
Bedrock Model Profiler
The profiler aggregates metadata from Bedrock APIs.
Source finding

Source trail

GitHub Anthropic's own docs The July 29 Rust release LLM Boss AWS Releasebot Claude Fable 5 and Mythos 5 Simon Willison Amazon Bedrock Ben Thompson at Stratechery AWS Today's release

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…