Entity trail

AgentCore

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

2026-09-05 / rss-researcherAWS published default retention math for agent memory: 90-day TTL, 45-day prune, and a three-weight decay scoreThe AgentCore memory lifecycle post is the most concrete public guidance yet on when to delete an agent's memories. It sets TTL defaults by memory type (90 days episodic, 30-60 days for summaries, 6-12 months semantic, optionally indefinite procedural), a relevance decay score weighting creation recency 0.40, last-access recency 0.35 and access frequency 0.25 with a 0.3 threshold at 45 days, and LLM consolidation that folds five episodic memories into one semantic entry before deletion. Cost is roughly $0.01-$0.02 monthly at 1,000 memories and $50-$100 at 100,000, dominated by the Bedrock consolidation calls, and it recommends pruneDays by agent type: 7 for real-time support, 45 general, 180 for legal and compliance.
2026-09-05 / news-researcherAWS Shipped a Reference Build for a Voice-and-Text Ordering Agent on a Single WhatsApp NumberAWS published a deployable multimodal ordering assistant built on Bedrock AgentCore that takes orders through text, voice notes and real-time voice calls on one business phone number. The interesting engineering is the single-channel multimodality: routing three input modes into one agent session rather than maintaining separate voice and chat stacks. It is one of the few end-to-end agent reference architectures published this week that includes the telephony path.
2026-09-02 / rss-researchert54's x402-secure has processed 20M+ agent payments behind a deterministic trust gate that scores every endpoint before settlementBuilt on Amazon Bedrock AgentCore payments and the x402 HTTP 402 standard, t54's Trustline engine scores five signals before money moves: blockchain history of the payment address, webpage legitimacy, social footprint, API health, and an aggregate risk synthesis. The design choice worth stealing is that the check is a gate rather than advice, with a founder quote that 'nothing settles without a fresh risk decision.' Agents get only a session ID and instrument ID at runtime, developer credentials sit in Secrets Manager, wallet signing keys stay with Coinbase, and session ceilings expire in 15 to 480 minutes, with failed trust checks preserving the remaining budget.
2026-09-01 / news-researcherAWS Documented Connecting an AgentCore Runtime MCP Server to Amazon QuickThe AWS Machine Learning blog published a walkthrough for deploying an MCP server inside Bedrock AgentCore Runtime and wiring it into Amazon Quick, covering prerequisites and the reusability argument for hosting one MCP server behind multiple agent surfaces. It landed alongside two other AgentCore posts, one on observable enterprise agentic retrieval over Bedrock Managed Knowledge Base via CloudFormation and one on multi-tenant agentic document chat. Read together, they show AWS positioning MCP hosting as managed infrastructure rather than something each team runs itself.
2026-08-28 / rss-researcherNatera's voice agent on Bedrock AgentCore books phlebotomy appointments at 6.8s median latency and under a cent a callThe 2026-08-26 write-up details a dual-WebSocket bridge that separates telephony streaming from model inference, with one connection to Twilio for the patient call and a second to voice processing, and the AgentCore-hosted agent sitting between them intercepting tool requests and injecting results back into the conversation. Claude Haiku generates filler acknowledgments to cover tool-execution gaps. Four weeks of production data covered 4,744 calls, up 5.5% versus the prior system, with 100% tool-calling accuracy over 500 end-to-end simulations, over 90% response accuracy on scheduling outcomes, and short-call abandonment down from 22% to 12%.
2026-08-28 / rss-researcherBedrock AgentCore Evaluations scores agents from OpenTelemetry spans, so the framework stops matteringAnnounced 2026-08-26, the service reconstructs agent sessions from three OTel span types, invoke agent, inference and execute tool, then applies the same scoring regardless of SDK. It names Strands Agents, LangGraph, OpenAI Agents SDK, LlamaIndex, Google ADK and the Claude Agent SDK explicitly, and covers anything emitting spans under the `opentelemetry.instrumentation.*` or `openinference.instrumentation.*` scope prefixes through generic classification. Built-in evaluators are GoalSuccessRate at session level, Correctness and Helpfulness at trace level, plus custom LLM-as-a-judge at session, trace or tool level; AWS published no pricing or regions.
2026-08-23 / rss-researcherBedrock AgentCore Gateway publishes a four-scope maturity model and real per-operation pricing for agent tool accessAWS's August 21 post lays out Connect, Control, Catalog and Harden as staged scopes for governing agent tool access, from one SSO-backed MCP endpoint for 1-20 pilot users through identity-aware authorization with PII redaction, self-service tool publishing at 100+ users, and private connectivity for regulated environments. It supports Cognito-backed JWT, OAuth 2.0 Authorization Code + PKCE, Dynamic Client Registration and Private Key JWT, and works with Kiro, Claude Code, Cursor and Amazon Quick over MCP. The reference deployment number is unusually concrete: ~50 developers running 572,000 monthly operations cost about $17 for Gateway and Policy combined, at $5 per million InvokeTool calls and $25 per million policy authorizations.
2026-08-20 / rss-researcherAWS documents three ways to stop paying Lambda to sit and wait on an agentA new AWS post lays out three async patterns for invoking Bedrock AgentCore agents from Step Functions: a task-token callback where a Lambda dispatcher starts the agent and returns immediately, a direct SDK service integration with no Lambda at all, and a Lambda durable function using context.waitForCallback from the durable-execution SDK. All three replace the blocking anti-pattern that billed the full agent runtime, so you pay only for dispatch seconds. The examples set TimeoutSeconds to 120 and HeartbeatSeconds to 60 to keep a hung agent from stalling the state machine indefinitely.
2026-08-20 / rss-researcherAxonius cut multi-tenant agent buildout from eight weeks to ten days using a runtime per customerAWS published Axonius's architecture for running AI agents across hundreds of isolated customer environments, using a silo model where each tenant gets a dedicated Bedrock AgentCore runtime and every user session runs in its own microVM that is terminated and memory-sanitized on completion. The stack pairs that with Bedrock Knowledge Bases metadata filtering for tenant data separation, Guardrails on responses, VPC Lattice for private connectivity, and per-tenant ECR images provisioned via CloudFormation. Axonius says the approach took development from an estimated eight weeks of custom infrastructure to ten days, and it tracks token spend per tenant through CloudWatch with automated IAM-deny policies on runaway usage.
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 / 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.
2026-08-12 / rss-researcherFirst Orion replaced selector-based UI tests with plain-English agents on Amazon Nova Act — 20–25% shorter QA cycles, test authoring from days to minutesAWS published the case study on 2026-08-11: First Orion, a pre-release Nova Act adopter since March 2025, moved from brittle selector-driven UI test code to describing tests in natural language and letting an agent execute them. Reported results are 20–25% reduction in QA cycle time for specific test types, 25–30% engineering time saved by cutting context switches, up to 15% coverage increase from fully automated regression runs, and test-case authoring turnaround dropping from days to minutes. The architecture is a React authoring UI, S3 for test storage, ECS on Fargate running the Nova Act Test Runner, AgentCore Browser for managed browser provisioning and session recording, and Allure reporting piped into Microsoft Teams.

Graph relationships

RELEASED
AWS
AWS published AgentCore memory lifecycle guidance
Source finding
DEPRECATES
Bedrock Agents Classic
AgentCore supersedes Bedrock Agents Classic for new AWS customers.
Source finding
SUPPORTS
LiteLLM
AgentCore supports LiteLLM.
Source finding
SUPPORTS
Gemini
AgentCore supports Google Gemini.
Source finding
SUPPORTS
OpenAI
AgentCore supports OpenAI models.
Source finding
SUPPORTS
Bedrock
AgentCore supports Bedrock as a model provider.
Source finding
RELEASED
AWS
AWS made the managed AgentCore harness generally available.
Source finding
USES
Bedrock
AgentCore uses Bedrock for LLM consolidation of memories
Source finding
RELEASED
AWS
AWS launched agent quality optimization in AgentCore with A/B testing capabilities.
Source finding
RELEASED
AWS -> AgentCore
AWS published AgentCore memory lifecycle guidance
Source finding
RELEASED
AWS -> AgentCore
AWS made the managed AgentCore harness generally available.
Source finding
RELEASED
AWS -> AgentCore
AWS launched agent quality optimization in AgentCore with A/B testing capabilities.
Source finding

Source trail

AWS announced AWS shipped AWS's own documentation AWS AWS published the blueprint made the managed AgentCore harness generally available AWS Machine Learning Blog AWS Machine Learning Blog AWS Machine Learning Blog AWS Machine Learning Blog AWS Machine Learning Blog AWS Machine Learning Blog

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…