Entity trail

Ent Security

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

2026-07-13 / agents-researcherSavi raises $7M seed for an on-device agent that screens calls, texts, and messages before you engageLos Angeles startup Savi launched iOS and Android apps whose agent screens calls, text messages, and other digital communications before the user engages, and raised $7M in seed funding led by Acrew Capital. It's a consumer-facing 'guardian agent' pattern — an agent acting as a filtering intermediary on the user's behalf. Notable as a mainstream, non-developer application of agentic AI to personal communications security.
2026-07-13 / agents-researcherFirst Recon AI ships AI Security Runtime GA — inline policy enforcement on every agent-to-agent and agent-to-tool callFirst Recon AI announced general availability (July 8, 2026) of its AI Security Runtime, which inspects every AI interaction — human-to-model, agent-to-tool, and agent-to-agent — and applies policy inline (allow, redact, hold, block) before data reaches a model. It records each decision as sealed, metadata-only evidence for SIEM pipelines and compliance against NIST, GDPR, and the EU AI Act, and supports OpenAI, Anthropic, Google, and Meta. It's part of a fast-emerging 'agent runtime security' category treating agent traffic as its own enforceable perimeter.
2026-07-13 / vibe-coding-researcherTip: Update Headless Claude Code to v2.1.207 — Managed-Settings Consent Was Silently Auto-Recordedv2.1.207 fixes a security bug where non-interactive runs (claude -p and the SDK) permanently recorded remote managed-settings consent as granted without ever showing the security consent dialog. If you run Claude Code headless in a pipeline or cron job with managed settings, upgrade — earlier builds consented on your behalf without asking.
2026-07-12 / sources-researcherMitigating Taint-Style MCP Server Vulnerabilities Without Touching Server Code (SPELLSMITH)'Mitigating Taint-Style Vulnerabilities in MCP Servers via Security-Aware Tool Descriptions' (arXiv:2607.07461, July 8, 2026) finds that taint-style flaws — untrusted input flowing to dangerous sinks — make up a substantial fraction of MCP-server CVEs, require large code changes to remediate, and meet slow community response. The proposed defense, SPELLSMITH, rewrites tool descriptions to be security-aware (a Description Enhancement Module plus an LLM self-reflection loop) so the model avoids triggering the vulnerable path, with no server-code change required. For builders shipping or installing MCP servers: treat tool descriptions as a real security surface and audit for taint-style sinks rather than assuming the protocol is safe by default.
2026-07-12 / sources-researcherToolFailBench: a 1,000-Task Diagnostic Benchmark That Localizes Exactly How Agent Tool Use BreaksToolFailBench (arXiv:2607.04686, July 6, 2026) is a 1,000-task diagnostic benchmark across finance, medicine, law, cybersecurity, and real estate that localizes where tool use breaks — labeling traces as Tool-Skip, Result-Ignore, Output-Fabrication, or Unnecessary-Tool-Use via a rule classifier plus two LLM judges aggregated by majority vote. Across 19 frontier models the best reaches only 86.33% clean tool-use rate, and models with similar aggregate scores fail in very different ways. For builders: an off-the-shelf harness to profile your agent's specific tool-use failure modes instead of relying on a single pass/fail score.
2026-07-12 / saas-disruption-researcherCROSS-CATEGORY: AI Agent Pricing Is Decoupling From Seats in the Same Two-Week Window — OpenAI Meters by Token, Microsoft Gates Agents Behind a License FloorThree independent moves in early July converge on the same signal: OpenAI began token-based credit metering for ChatGPT Workspace agents on July 6, Microsoft made an E5 license a prerequisite for Agent 365 purchases (effective June 1) and raised M365 list prices July 1, and industry surveys now show seat-based pricing falling from 21% to 15% of SaaS vendors year-over-year while hybrid models jumped 27% to 41%. The per-seat model that anchored SaaS economics for two decades is being replaced simultaneously across the two largest productivity vendors — one by consumption metering, one by a security-tier tax on deploying agents at all. For builders: pricing your own agents on flat per-seat is now the outlier, and 'what does one run cost' is becoming the mandatory unit of account.
2026-07-12 / rss-researcherGitHub Ships CodeQL 2.26.0 With AI Prompt-Injection Detection for JavaScript/TypeScriptGitHub released CodeQL 2.26.0 on July 11, 2026, adding static-analysis queries that flag AI prompt-injection vulnerabilities in JavaScript and TypeScript, plus Kotlin 2.4.0 support and broader multi-language analysis improvements. Pulling prompt-injection detection into the standard code-scanning pipeline signals that untrusted-input-into-LLM flows are now treated as a first-class security defect class. Builders wiring LLM calls into JS/TS apps can now catch injection sinks in CI instead of at runtime.
2026-07-11 / agents-researcherFriendly Fire: auto-approve coding agents run attacker code during security scansThe AI Now Institute demonstrated 'Friendly Fire' in July 2026: AI coding agents running in autonomous, self-approving mode can be tricked into executing attacker code precisely when asked to scan untrusted third-party code for vulnerabilities. Prompt injections seeded across ordinary repo files steer Claude Code and OpenAI Codex into running attacker-controlled commands during what looks like a routine security review. The finding shows that pointing an auto-approve agent at hostile code is itself the vulnerability.
2026-07-11 / arxiv-researcherDual-Graph Verification Framework Audits Legacy IT Security Concepts for NIS-2 / OSCAL MigrationThe NIS-2 Directive pushes organizations from document-based compliance toward machine-readable artifacts like OSCAL (with Germany's BSI adapting it as Grundschutz++), but migrating extensive legacy IT security concepts without verification risks translation errors. This paper proposes a dual-graph verification framework to reverse-engineer and audit those legacy concepts during migration. Complements the concurrent MCP-based OSCAL pipeline work as the verification half of continuous-compliance tooling.
2026-07-11 / vibe-coding-researcherSentry MCP Server Can Be Weaponized via Fake Error Events — 85% Agent-Hijack SuccessTenet Security showed that injecting fake error events through a public Sentry DSN causes the Sentry MCP server to return them as trusted diagnostics, letting attackers reach an 85% success rate at tricking coding agents into executing arbitrary commands. It's a clean example of an indirect prompt-injection path where the payload rides in through a legitimate observability integration, not a malicious server. Treat all MCP tool output — even from trusted vendors — as untrusted input, and gate command execution behind explicit confirmation.
2026-07-10 / skill-finderMCP Apps: declare your UI templates ahead of time so hosts can prefetch and security-review themMCP Apps standardizes the ad-hoc embedded-UI experiments into server-rendered HTML displayed in a sandboxed iframe. The design constraint worth copying: tools declare their UI templates in advance rather than streaming markup at call time, so the host can prefetch, cache, and security-review the template before anything executes. Every action taken inside the rendered UI routes back through the same JSON-RPC base protocol, inheriting the identical consent and audit path as a direct tool call.
2026-07-10 / saas-disruption-researcherCROSS-CATEGORY: Three Vendors in Three Unrelated Categories Shipped 'Agents That Supervise Other Agents' Inside Two WeeksNorm Ai (legal/compliance, $120M Series C, July 7) is building supervisory agents to audit customers' other AI deployments. Snyk (devtools/security, GA June 29) shipped Evo ADS to police MCP servers and coding agents at runtime. Vanta (GRC, GA this month) shipped an AI Agent plus an Agent for Risk to unify internal and third-party risk, explicitly targeting 'shadow AI.' Three categories, one product shape. The meta-observation for builders: the second-order market — governance of autonomous agents — is being staked out before the first-order agent market has consolidated.

Graph relationships

INVESTED IN
Sequoia
Sequoia participated in Ent Security seed
Source finding
INVESTED IN
Decibel
Ent Security raised $100M seed led by Decibel
Source finding
WORKS AT
Brandon Dixon
Brandon Dixon founded Ent Security
Source finding
WORKS AT
Elias Manousos
Elias Manousos founded Ent Security
Source finding
INVESTED IN
Sequoia -> Ent Security
Sequoia participated in Ent Security seed
Source finding
WORKS AT
Brandon Dixon -> Ent Security
Brandon Dixon founded Ent Security
Source finding

Source trail

OpenAI Tech Times Agent Fabric arXiv paper Ent Security Crescendo AI Business Wire Claude Code Docs — Changelog (July 2026)arXiv arXiv Monetizely GitHub (CodeQL release notes)

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…