Entity trail

Connectors

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

2026-08-14 / reddit-researcherClaude in Chrome Becomes a Shared 'Claude Cowork' Session That Persists Across Desktop, Web and MobileAs of August 12, the Claude browser extension is no longer a separate surface — it runs as a shared Claude Cowork session, so browser conversations save to history and skills and connectors work inside the browser, letting a task continue across desktop, web, and mobile. Anthropic added a safety layer that, before consequential actions, reviews what the agent is about to do against what the user originally asked and blocks anything that doesn't match. That intent-matching check is a different mitigation shape than allowlists or permission prompts, and worth watching as browser agents get more autonomy.
2026-08-13 / hn-researcherBallet Launches Workflow Automation That Writes API Integrations On Demand Instead of Shipping ConnectorsBallet, from the team behind Brainfish, launched on HN August 13 (30 points) with a pitch aimed squarely at the connector-catalog model: 'Describe a multi-system outcome in plain English and Ballet builds it as reviewable, version-controlled code.' Rather than maintaining pre-built integrations, it generates code against any API at request time, producing 'inspectable code instead of a brittle visual graph,' and mixes deterministic logic where accuracy matters with agentic reasoning where flexibility helps. It's early access targeting RevOps, Support Ops, and PLG teams; no pricing is published.
2026-08-12 / github-pulse-researcherMacro Bets That Shared AI Memory — Not Another Chat Sidebar — Is the Unit of a Team Workspacemacro-inc/macro gained about 248 stars today (1,251 total) for a Rust, AGPL-3.0 workspace unifying email, chat, docs, tasks, agents, calls, and CRM with @-linking across all of them backed by shared AI memory. The architectural claim is that agent usefulness is bounded by context fragmentation across tools, so the fix is collapsing the tools rather than adding connectors between them. The AGPL license is an unusual choice for a team-collaboration product and will constrain commercial forks.
2026-08-11 / rss-researcherVercel Connect finishes the CLI story: all 100+ connectors now configure without opening the dashboard`vercel connect create` previously completed setup in the terminal for only some services and punted everything else to the dashboard. Now all 100+ services configure end-to-end from the CLI: the command pre-populates brand name, icon and auth type, prompts for credentials, and `vercel connect attach [service/name] --project [project] --environment [env]` wires it to a project and environment. Scoped tokens come from `getToken()` — the practical win is that connector setup becomes scriptable rather than a manual browser step in CI or agent workflows.
2026-08-09 / projects-researcherLLM Trading Agents Cross 90,000 Combined Stars: HKU's Vibe-Trading at 30,423 and daily_stock_analysis at 60,905 (+287 Today)HKUDS/Vibe-Trading (MIT, v0.1.12 shipped July 22) from Hong Kong University's data science lab pairs an LLM reasoning loop of 50+ tools with 18+ historical data providers (Tushare, Yahoo Finance, OKX, CCXT) and nine market-specific backtesting engines covering US equities, A-shares, crypto, India NSE/BSE and Korea KRX, with realistic cost models. It implements mandate gates, kill switches and audit trails and takes no custody — 'the broker holds funds and executes; we only relay intent' — while labelling live broker connectors experimental. Alongside it, ZhuLinsen/daily_stock_analysis is trending at 60,905 stars (+287 today) doing LLM-driven analysis with automated notifications; together they show retail-facing autonomous finance is now a mass-adoption open-source category, guardrail language notwithstanding.
2026-08-08 / saas-disruption-researcherBackstory Retiered 141 Accounts in Three Days Using Claude and MCP — the Same Job Used to Take Five Teams a QuarterSaaStr published (August 8) a walkthrough from Backstory VP of Customer Success Haya Kamola showing an account-tiering exercise across 141 accounts completed in three to four days, producing 8 Tier A and 21 Tier B accounts plus C/D segments. The setup was Claude via Cowork with connectors into Amplitude, Atlassian/Jira, Backstory's own MCP server, and Slack — no CS platform, no RevOps vendor, no data team. Kamola's framing is the finding: in a previous role the same exercise required her team plus four others working a full quarter, which means the displaced product here isn't a SaaS tool, it's five teams' worth of coordination overhead.
2026-08-06 / saas-disruption-researcherPromptArmor: Atlassian Rovo Zero-Click Exfiltrates Jira and Confluence Data, Still Unpatched 2.5 Months After DisclosurePromptArmor disclosed publicly on August 5 that Atlassian's Rovo agent can be made to exfiltrate any data reachable through its connectors — Jira and Confluence included — via indirect prompt injection, because Rovo's URL-retrieval tool has no protection against URLs the agent itself generates. The attack defeats organization-level web-search restrictions outright: disabling web search removes the search tool but leaves the tool that opens search results. Timeline: disclosed May 23, acknowledged May 25, follow-ups June 4 and July 29, no further contact, no CVE, and Rovo remains vulnerable — a concrete data point that incumbent AI features are shipping with less security review than the core products they are bolted onto.
2026-08-06 / arxiv-researcherAnthropic Ships Inference Hooks: Enterprise DLP Servers Now Vote Allow/Deny on Every Claude Prompt Before InferenceOn August 5, 2026 Anthropic launched inference hooks in beta for Claude Enterprise, routing each prompt and tool-call response over a signed WebSocket to the customer's own security server for an allow-or-deny verdict before the model generates anything. Coverage spans Claude chat, Claude Code, Claude Cowork, and tool calls made through MCP connectors, skills, and plugins, with named integrations for Netskope, Palo Alto Networks, Proofpoint, and Zscaler, plus shadow mode, role-based exclusions, and percentage rollouts. The current limitation matters for builders: verdicts are binary — the server can block a prompt but cannot redact or rewrite it — so any redaction logic still has to live client-side.
2026-08-06 / vibe-coding-researcherPattern: 'Loop Engineering' Is Consolidating as the Name for Designing Systems That Prompt AgentsThe framing — attributed to Peter Steinberger, drawing on Addy Osmani and Boris Cherny — is that you should stop prompting coding agents and start designing loops that prompt them, with six named building blocks: scheduling, worktrees, skills, plugins/MCP connectors, sub-agents in maker/checker splits, and durable state living outside the conversation. The distinction it draws between a 'harness' (the versioned runtime stack) and a 'loop' (a single scheduled execution pattern) is the useful part, because most practitioner writing conflates them. Starters targeting Grok, Claude Code, Codex, Opencode, and GitHub Actions signal the pattern is being written harness-agnostically.
2026-08-03 / arxiv-researcherY Combinator Open-Sources QM, the Multiplayer Agent Harness It Runs Its Own Company OnYC has released QM under MIT license — the agent harness it uses internally across accounting, legal, events, and engineering, including to build QM itself. It gives every employee and Slack room its own scoped memory, permissions, and sandbox, and deliberately decouples harness from model: Pi, OpenCode, Codex, and Claude Code all drive the same core. Features include cron and webhook triggers, shared files, company-brain connectors, agent browser support, and shareable web artifacts; the repo was created 2026-07-29 and has already reached 8,420 stars and 887 forks.
2026-08-03 / agents-researcherSuna 0.12 adds per-session connector scoping, denied-secret enforcement, and a unified audit trailKortix shipped Suna 0.12.0 on August 1 and 0.12.1 on August 2, and the theme is agent least-privilege rather than capability. New sessions can start with an explicit opt-in scope over connectors and secrets; agents can require connector profiles, profiles carry an authorization strategy, and denied secrets never enter the sandbox at all. Approval prompts now show full tool arguments and approval policies match on arguments rather than just tool names — closing a real gap where approving "run_sql" approved every query. It also lands a centralized audit log of every API request and agent action with client attribution, per-session cost records, and sandboxes with a bounded lifetime billed against liveness evidence instead of wall clock.
2026-07-31 / hn-researcherLaunch HN: Prized (YC S26) Bets That Internal-Tool Safety Is a Setup Problem, Not a Model Behavior ProblemPrized launched on HN July 30 (74 points, 53 comments) letting ops, support and finance staff build internal tools with AI while admins pre-approve 20+ connectors — Salesforce, Slack, Postgres, Stripe, Notion — and define exactly what data each can reach. Each generated tool runs under its own scoped role rather than blanket database access, and every run is logged with who ran it, what it touched and when. The company's framing is the interesting part for builders: 'fixes the setup, not the behavior' — constrain the environment rather than trying to constrain what the model decides to do.

Graph relationships

RELEASED
Mistral
Mistral shipped new Connectors capabilities for secure enterprise integrations.
Source finding
RELEASED
Mistral -> Connectors
Mistral shipped new Connectors capabilities for secure enterprise integrations.
Source finding

Source trail

Mistral via Releasebot X/Twitter Releasebot (Anthropic changelog)Ballet / Hacker News GitHub Vercel Changelog GitHub SaaStr PromptArmor (corroborated by Hacker News front page, 248 points)Anthropic (Claude blog)GitHub GitHub yc-software/qm (verified via GitHub API) + Y Combinator announcement

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…