★ NO REAL MONEY ★ Will an AI browser agent be shown exfiltrating user data via prompt in — YES 70.0% ◆ INTEL: Open WebUI Desktop - native app for running local or remote LLMs ◆ INTEL: Announcing OSV-Scanner V2: Vulnerability scanner and remediation tool for open s ◆ Will a prompt injection attack be publicly blamed for a data breach at — YES 62.0% ◆ INTEL: RedAI - AI-driven vulnerability discovery and live validation ◆ Which AI lab discloses the next headline security incident? — OpenAI 46.2% ◆ INTEL: Malware in Open VSX: These Vibes Are Off - extension risks for AI code editors ◆ Will a jailbreak prompt sell as an NFT for more than 1 ETH before 2027 — NO 95.0% ◆ Will an executive literally say 'it's priced in' about a breach on an — NO 85.0% ◆ Will itspricedin.lol itself get pwned within 90 days of launch? — YES 99.1% ◆ Will a CVSS 9.0+ RCE be disclosed in a major AI coding agent before 20 — YES 80.5% ◆ Will $POC ever be worth anything? — NO 99.9% ◆ INTEL: Red Team Ops course - Zero-Point Security ◆ Will a major password manager disclose a CVSS 9.0+ vulnerability befor — NO 60.0% ◆ Will a leaked system prompt lead to a public apology from a major AI l — NO 70.0% ◆ INTEL: ADPulse - Active Directory security scanner ◆ Will a global Cloudflare, AWS, or Azure outage lasting over 3 hours be — YES 55.0% ◆ INTEL: PolicyEval - policy compliance evaluation and enforcement for LLM outputs ◆ INTEL: Ross Hendricks on AI as mass psychosis: zero productivity gains after $3 trillio ◆ Will AI-generated bug reports cause a major open-source project to shu — YES 55.0% ◆ ★ $POC HAS NO VALUE ★ ◆
🕵️ Intel feed — 5660 resources from the ai-security compendium
Clear

Categories: Agent Frameworks · Agent Identity · Agent Security · Anti-Crawling · Architecture · Benchmarks · Code Analysis · Coding Tools · Detection and Monitoring · Engineering Patterns · Frameworks and Standards · General Reading · Governance and Policy · Guardrails and Firewalls · Honeypots and Deception · Incidents · Jailbreaking · MCP (Model Context Protocol) · Model Attacks · Papers · Privacy · Prompt Injection · Red Teaming · Safety and Alignment · Sandboxing and Isolation · Secrets Management · Secure Coding · Supply Chain · Threat Modeling

Showing 48 of 48. Source: github.com/AnthonyHerman/ai-security-corpus · browse it in Internet Explorer. This is the research; the markets are the bets.

Snapshot 2026-10-10 · last checked 2026-10-13 14:25 UTC · NEW = added in the last 14 days (nothing fresh this fortnight; the bugs are resting)

TitleCategorySection
Building Effective AI Coding Agents for the Terminal: Scaffolding, Harness, Context Engineering, and Lessons LearnedEngineering PatternsHarness Engineering
The Anatomy of an Agent HarnessEngineering PatternsHarness Engineering
Harness engineering: leveraging Codex in an agent-first worldEngineering PatternsHarness Engineering
Harness Engineering - Martin FowlerEngineering PatternsHarness Engineering
The Emerging 'Harness Engineering' PlaybookEngineering PatternsHarness Engineering
Skill Issue: Harness Engineering for Coding AgentsEngineering PatternsHarness Engineering
Harness Design for Long-Running Application DevelopmentEngineering PatternsHarness Engineering
Better Harness: A Recipe for Harness Hill-Climbing with EvalsEngineering PatternsHarness Engineering
Addy Osmani on Loop Engineering - designing coding agent systemsEngineering PatternsHarness Engineering
Adversarial Review: Structured Disagreement for Grounded Agentic Code ReviewEngineering PatternsHarness Engineering
Agent-to-agent pair programming - letting Claude and Codex work together as worker and reviewerEngineering PatternsHarness Engineering
Agentic Context Management: Solving Agent Memory and Cost as Lifecycle and Architecture ProblemsEngineering PatternsHarness Engineering
Agentic Harness EngineeringEngineering PatternsHarness Engineering
Agentic Harness Engineering - observability-driven evolution of coding-agent harnessesEngineering PatternsHarness Engineering
Agentic Harness Engineering: Observability-Driven Automatic Evolution of Coding-Agent HarnessesEngineering PatternsHarness Engineering
Andy Konwinski introduces Headlong, an open source microharness for persistent agentsEngineering PatternsHarness Engineering
Autoloop - GitHub Next goal-driven agent loop for automating tasksEngineering PatternsHarness Engineering
Boris Cherny: every 6 months delete your CLAUDE.md, skills and hooks and see what the model doesEngineering PatternsHarness Engineering
Build your own vulnerability harnessEngineering PatternsHarness Engineering
Building a Personal AI Factory - July 2025 snapshotEngineering PatternsHarness Engineering
Claude Code creator writes loops instead of promptingEngineering PatternsHarness Engineering
Explore the context window - Claude Code DocsEngineering PatternsHarness Engineering
Harness Engineering - ai-literacy-superpowers explanationEngineering PatternsHarness Engineering
Harness Engineering for Self-Improvement - Lilian WengEngineering PatternsHarness Engineering
improve-codebase-architecture - mattpocock Claude Code skillEngineering PatternsHarness Engineering
Learn Harness Engineering - project-based course for reliable Codex and Claude CodeEngineering PatternsHarness Engineering
Natural-Language Agent HarnessesEngineering PatternsHarness Engineering
OpenAI Symphony spec - orchestrating isolated autonomous coding agent runsEngineering PatternsHarness Engineering
Personal AI Infrastructure v5.0.0 release notes - Daniel MiesslerEngineering PatternsHarness Engineering
prove_it - verification harness for Claude Code that forces tests to run via hooksEngineering PatternsHarness Engineering
Ralph Wiggum as a "software engineer"Engineering PatternsHarness Engineering
Structural Backpressure Beats Smarter AgentsEngineering PatternsHarness Engineering
Thariq - Lessons from Building Claude Code: How We Use SkillsEngineering PatternsHarness Engineering
The advisor strategy: Give Sonnet an intelligence boost with OpusEngineering PatternsHarness Engineering
The Bitter Lesson of Tool CallingEngineering PatternsHarness Engineering
Welcome to Gas Town - Steve Yegge on orchestrating fleets of Claude Code agentsEngineering PatternsHarness Engineering
What Changes When the Atomic Unit of Intelligence Is No Longer a Single LLM CallEngineering PatternsHarness Engineering
Why We Built Our Own Background Agent - Ramp BuildersEngineering PatternsHarness Engineering
Your Agent Harness Is Not an Orchestration LayerEngineering PatternsHarness Engineering
A Sufficiently Detailed Spec Is CodeEngineering PatternsGeneral Insights
Addy Osmani on The Orchestration Tax - human attention as the bottleneck for multi-agent workflowsEngineering PatternsGeneral Insights
Andrej Karpathy's notes from Claude coding: agent workflows, fallibility, leverage, slopacolypseEngineering PatternsGeneral Insights
How to Fix Your ContextEngineering PatternsGeneral Insights
I Can't Do That, Dave - No Agent Yet: why AI agents need identity and the ability to say noEngineering PatternsGeneral Insights
Jonathan Vanderford on AI agents driving a return to detailed specs and upfront designEngineering PatternsGeneral Insights
Learnings from building AI agents - how cubic made its AI code reviewer less noisyEngineering PatternsGeneral Insights
The Pipeline Is Dead, Long Live the Agent MeshEngineering PatternsGeneral Insights
Thoughts on slowing the fuck down - how we should work with coding agentsEngineering PatternsGeneral Insights
Past performance is not indicative of future CVEs.
📎
💰 get $POC --:--