Shareable cards — Monday, July 13, 2026

Monday, July 13, 2026 — top 10 database + AI stories, one card each. Save the image, copy its caption, post.

Phone: tap Share image, pick Save Image in the sheet (LinkedIn's app won't take images from Safari directly), then tap the button again — it opens LinkedIn: start a post, add the photo on top, paste the caption. Desktop: Open → save the image, attach it in the composer, paste the caption.

Hand-drawn sketchnote — GhostLock: 15-year-old stack-UAF vulnerability in Linux kernel

Open image

GhostLock: 15-year-old stack-UAF vulnerability in Linux kernel

The day's edition as one sketchnote — full stories + sources:
https://offnetchat.com/news/2026-07-13.html?s=db926932f891e1a3#item-db926932f891e1a3

#TechNews #AI

PoWA Archivist 5.2.0 Adds PostgreSQL 19 Support

Open image

PoWA Archivist 5.2.0 Adds PostgreSQL 19 Support

▸ PoWA now supports PostgreSQL 19, essential for early adopters of the latest release.
▸ New pg_stat_recovery and pg_stat_lock views enable deeper internal diagnostics.

Summary + source: https://offnetchat.com/news/2026-07-13/s/d2903a018bbb6d1c.html

#Databases #TechNews

RegreSQL 1.0 launches to detect query performance regressions using production statistics

Open image

RegreSQL 1.0 launches to detect query performance regressions using production statistics

▸ RegreSQL 1.0 shifts focus from row correctness to execution plan stability.
▸ Uses production statistics to simulate real-world query planning behavior.

Summary + source: https://offnetchat.com/news/2026-07-13/s/cd15556c2ba4a9c2.html

#Databases #TechNews

GATS replaces LLM inference with layered world models for efficient agent planning

Open image

GATS replaces LLM inference with layered world models for efficient agent planning

▸ GATS removes LLM inference during the planning phase, significantly cutting computational costs.
▸ A layered world model handles known actions via symbolic matching and logs, reserving LLMs for unknowns.

Summary + source: https://offnetchat.com/news/2026-07-13/s/9ef86e16f82bfe26.html

#AgenticAI #TechNews

KV-PRM cuts Process Reward Model latency by reusing KV caches

Open image

KV-PRM cuts Process Reward Model latency by reusing KV caches

▸ Eliminates quadratic scoring cost by reusing existing KV caches instead of re-encoding text.
▸ Enables efficient Process Reward Modeling for long-context multi-agent rollouts.

Summary + source: https://offnetchat.com/news/2026-07-13/s/cd930412c5ac3c55.html

#AgenticAI #TechNews

Shared Selective Persistent Memory for Agentic LLM Systems

Open image

Shared Selective Persistent Memory for Agentic LLM Systems

▸ Agentic LLMs lose productive context like schemas and constraints between sessions, hurting efficiency.
▸ Selective persistence keeps reusable artifacts (specs, schemas) while dropping verbose reasoning traces.

Summary + source: https://offnetchat.com/news/2026-07-13/s/a301014cfb00c4cd.html

#AgenticAI #TechNews

Ploy reports 2.2x speed and 27% cost drop moving production AI agent to GPT-5.6

Open image

Ploy reports 2.2x speed and 27% cost drop moving production AI agent to GPT-5.6

▸ GPT-5.6 delivers 2.2x faster inference times compared to the prior model version in production.
▸ Migration yields a 27% reduction in costs, improving the unit economics of AI agent operations.

Summary + source: https://offnetchat.com/news/2026-07-13/s/4f4c2bd39a3c173f.html

#AgenticAI #TechNews

Destructive Command Guard blocks risky shell commands in AI coding agents

Open image

Destructive Command Guard blocks risky shell commands in AI coding agents

▸ Prevents accidental data loss by blocking destructive shell commands in AI coding workflows
▸ Supports Claude Code, Cursor, Gemini CLI, Copilot, and others via hooks or compatibility layers

Summary + source: https://offnetchat.com/news/2026-07-13/s/3f7f2e4444aa6685.html

#AgenticAI #TechNews

CogniConsole externalizes LLM inference control to reduce failures

Open image

CogniConsole externalizes LLM inference control to reduce failures

▸ Reliability is driven by inference-time control layers, not just model size or capability.
▸ CogniConsole externalizes control via a structured interface for task framing and context.

Summary + source: https://offnetchat.com/news/2026-07-13/s/8d75ea8a3d7c27ec.html

#LLM #TechNews

GRACE: Graph-Regularized Context Evolution for Stable LLM Agents

Open image

GRACE: Graph-Regularized Context Evolution for Stable LLM Agents

▸ Replaces fragile flat-text instruction maintenance with a structured typed semantic graph.
▸ Validates updates locally to reduce verification complexity as context scales.

Summary + source: https://offnetchat.com/news/2026-07-13/s/60bfdd086712290e.html

#AgenticAI #TechNews

OpenProver: Agentic Lean 4 Theorem Proving with Human Steering

Open image

OpenProver: Agentic Lean 4 Theorem Proving with Human Steering

▸ Uses a Planner-Worker-Verifier architecture for structured, parallel proof generation.
▸ Integrates Lean 4 for automatic formal verification of all generated proofs.

Summary + source: https://offnetchat.com/news/2026-07-13/s/e2f300c146fd8b9d.html

#AgenticAI #TechNews