DestructoR666 AI Digest · Free-source v0.3 no-repeat edit

Signal
Rag

Jul 3, 2026 80 leads scanned 11 sources
A tiny anti-slop zine for agents, tools, model ops, and actually useful AI weirdness. No X API toll paid.

Editor picks

Thinking, experiments, papers, and opinionated field notes first. Releases are useful, but they live downstairs now.
#01
Hacker News · hn · Jul 3

Show HN: Scopewalker, an MCP server for codebase complexity metrics

Editor's read

A community-submitted project/thread. A title-first lead about: Scopewalker, an MCP server for codebase complexity metrics. Skim it only if that maps to current agent/tooling work. Treat it as an early smoke signal, not a verdict. HN: 1 points, 0 comments. Query: AI agents tool use.

Why you might care

MCP is becoming the plumbing layer for agent tools. Worth watching because it shapes what Hermes/OAC can connect to without bespoke glue code.

Source excerpt

HN: 1 points, 0 comments. Query: AI agents tool use.

hnAI agents tool use
read65.8agent, agents, ai agent, mcp
#02
Hacker News · hn · Jul 2

Launch HN: Manufact (YC S25) – MCP Cloud

Editor's read

A community-submitted project/thread. A title-first lead about: Launch HN: Manufact (YC S25) – MCP Cloud. Skim it only if that maps to current agent/tooling work. Treat it as an early smoke signal, not a verdict. HN: 108 points, 63 comments. Query: AI agents tool use.

Why you might care

MCP is becoming the plumbing layer for agent tools. Worth watching because it shapes what Hermes/OAC can connect to without bespoke glue code.

Source excerpt

HN: 108 points, 63 comments. Query: AI agents tool use.

hnAI agents tool use
read64.5agent, agents, ai agent, mcp
#03
Simon Willison · feed · Jul 2

llm-coding-agent 0.1a0

Editor's read

A post from Simon Willison. Plain version: Release: llm-coding-agent 0.1a0 Another Fable 5 experiment.

Why you might care

Kept because it matched the radar’s AI/agent/tooling filters. Skim title + source first; open only if it lines up with current Hermes/OAC work.

Source excerpt

Release: llm-coding-agent 0.1a0 Another Fable 5 experiment. Now that my LLM library has evolved into more of an agent framework it's time to see what a simple coding agent would look like built on it. I started a new Python library using my python-lib-template-repository GitHub template repository, then ran these two prompts (here's the Claude Code for web…

llmtoolspractical
read58.8agent, coding agent, claude code
#04
Simon Willison · feed · Jul 2

Understand to participate

Editor's read

A post from Simon Willison. Plain version: I saw Geoffrey Litt speak at AIE yesterday, and one framing he used particularly resonated with me: Understand to participate Geoffrey was talking about the challenge of collaborating with coding agents as the…

Why you might care

Kept because it matched the radar’s AI/agent/tooling filters. Skim title + source first; open only if it lines up with current Hermes/OAC work.

Source excerpt

I saw Geoffrey Litt speak at AIE yesterday, and one framing he used particularly resonated with me: Understand to participate Geoffrey was talking about the challenge of collaborating with coding agents as they construct increasingly large and sophisticated changes, and the need to avoid taking on cognitive debt as your understanding drifts from how the cod…

llmtoolspractical
read58.7agent, agents, coding agent
#05
Hacker News · hn · Jul 2

Show HN: Enola-A deterministic architecture graph for developers and AI agents

Editor's read

A community-submitted project/thread. A title-first lead about: Enola-A deterministic architecture graph for developers and AI agents. Skim it only if that maps to current agent/tooling work. Treat it as an early smoke signal, not a verdict. HN: 9 points, 4 comments. Query: MCP agent LLM.

Why you might care

MCP is becoming the plumbing layer for agent tools. Worth watching because it shapes what Hermes/OAC can connect to without bespoke glue code.

Source excerpt

HN: 9 points, 4 comments. Query: MCP agent LLM.

hnMCP agent LLM
read57.5agent, agents, ai agent, mcp

Articles / research / opinions

The stuff people are thinking, trying, arguing, and building with the tech. This is the main read pile.
#06
Simon Willison · feed · Jul 2

Using DSPy to evaluate and improve Datasette Agent's SQL system prompts

Editor's read

A post from Simon Willison. Plain version: Research: Using DSPy to evaluate and improve Datasette Agent's SQL system prompts One of this morning's AIE keynotes covered dspy , which reminded me I've been meaning to see if it could help me improve the sy…

Why you might care

Evals are how we avoid vibes-only agent work. Useful if it suggests tests we can adapt for Hermes/OAC reliability.

Source excerpt

Research: Using DSPy to evaluate and improve Datasette Agent's SQL system prompts One of this morning's AIE keynotes covered dspy , which reminded me I've been meaning to see if it could help me improve the system prompt used by Datasette Agent - so I fired off an asynchronous research task in Claude Code for web using Claude Fable 5: Pip install the latest…

llmtoolspractical
read55.8agent, eval, claude code
#07
Latent Space · feed · Jul 3

AIEWF Daily Dispatch: The great loops debate and the state of AI engineering

Editor's read

A post from Latent Space. Plain version: The AI Engineer World’s Fair ended with a debate about loops, a report on the state of AI engineering, and closing keynotes focused on what to build next.

Why you might care

Kept because it matched the radar’s AI/agent/tooling filters. Skim title + source first; open only if it lines up with current Hermes/OAC work.

Source excerpt

The AI Engineer World’s Fair ended with a debate about loops, a report on the state of AI engineering, and closing keynotes focused on what to build next.

aiindustryagents
read52.6agents, ai engineering
#08
Latent Space · feed · Jul 2

The website of the future may assemble itself for every visitor

Editor's read

A post from Latent Space. Plain version: Adobe is experimenting with “agentic sites” that generate pages around an individual user’s intent.

Why you might care

Kept because it matched the radar’s AI/agent/tooling filters. Skim title + source first; open only if it lines up with current Hermes/OAC work.

Source excerpt

Adobe is experimenting with “agentic sites” that generate pages around an individual user’s intent. At AIEWF, we talked to Carlos Sanchez about the Web's future.

aiindustryagents
read52.0agent, agents
#09
arXiv · paper · Jul 2

WorldDirector: Building Controllable World Simulators with Persistent Dynamic Memory

Editor's read

A research paper. In human terms: We present WorldDirector, a highly controllable video world model framework designed for persistent dynamic object memory and unrestricted viewpoint exploration.

Why you might care

This maps directly onto OAC: what agents remember, what they should forget, and how not to build a cursed context landfill.

Source excerpt

We present WorldDirector, a highly controllable video world model framework designed for persistent dynamic object memory and unrestricted viewpoint exploration. Unlike existing world models that entangle physical dynamics with pixel rendering and rely on continuous visual observation to sustain motion, our framework explicitly decouples semantic motion orchestration from visual generation. By leveraging an LLM to c…

paperLLM evaluation benchmark agentscs.CV
skim abstract45.7agent, agents, memory, eval
#10
arXiv · paper · Jul 2

Distributed Attacks in Persistent-State AI Control

Editor's read

A research paper. In human terms: As AI coding agents become more autonomous, they increasingly ship code iteratively, with the codebase persisting across sessions.

Why you might care

Evals are how we avoid vibes-only agent work. Useful if it suggests tests we can adapt for Hermes/OAC reliability.

Source excerpt

As AI coding agents become more autonomous, they increasingly ship code iteratively, with the codebase persisting across sessions. This persistence creates a new attack surface: a misaligned or prompt-injected agent can distribute attacks across pull requests (PRs) and time its payload for the PR with the best natural cover. To study the resulting dynamics, we introduce Iterative VibeCoding, a setting for AI control…

paperLLM evaluation benchmark agentscs.AI
skim abstract45.7agent, agents, coding agent, eval
#11
arXiv · paper · Jul 2

LACUNA: A Testbed for Evaluating Localization Precision for LLM Unlearning

Editor's read

A research paper. In human terms: LLMs memorize sensitive training data, including personally identifiable information (PII), creating a pressing need for reliable post hoc removal methods.

Why you might care

This maps directly onto OAC: what agents remember, what they should forget, and how not to build a cursed context landfill.

Source excerpt

LLMs memorize sensitive training data, including personally identifiable information (PII), creating a pressing need for reliable post hoc removal methods. Unlearning has emerged as a promising solution, with state-of-the-art(SOTA) methods often following a localize-first, unlearn-second paradigm that targets specific model parameters. However, existing benchmarks evaluate unlearning solely at the output level, leav…

paperlarge language model agents memorycs.CLcs.AI
skim abstract45.7agent, agents, memory, eval
#12
Latent Space · feed · Jul 2

Vercel's Andrew Qu on why agents are a new kind of software

Editor's read

A post from Latent Space. Plain version: The Vercel Chief of Software explains how its agent framework, eve, was created — and why skills, sandboxes and agent-readable websites now matter.

Why you might care

Kept because it matched the radar’s AI/agent/tooling filters. Skim title + source first; open only if it lines up with current Hermes/OAC work.

Source excerpt

The Vercel Chief of Software explains how its agent framework, eve, was created — and why skills, sandboxes and agent-readable websites now matter.

aiindustryagents
read45.2agents
#13
OpenAI News · feed · Jun 29

Inside Genebench-Pro

Editor's read

A post from OpenAI News. Plain version: A title-first lead about: Inside Genebench-Pro. Skim it only if that maps to current agent/tooling work.

Why you might care

Kept because it matched the radar’s AI/agent/tooling filters. Skim title + source first; open only if it lines up with current Hermes/OAC work.

Source excerpt

No summary from source. Open the link if this smells interesting.

openaimodelsagents
skim39.8agents
#14
OpenAI News · feed · Jun 29

Core dump epidemiology: fixing an 18-year-old bug

Editor's read

A post from OpenAI News. Plain version: OpenAI engineers used large-scale core dump analysis to debug rare infrastructure crashes, uncovering both a hardware fault and a long-standing software bug.

Why you might care

Kept because it matched the radar’s AI/agent/tooling filters. Skim title + source first; open only if it lines up with current Hermes/OAC work.

Source excerpt

OpenAI engineers used large-scale core dump analysis to debug rare infrastructure crashes, uncovering both a hardware fault and a long-standing software bug.

openaimodelsagents
skim39.8agents

Releases / software updates

Important for staying current, but intentionally below the longform material. Changelogs are seasoning, not dinner.
#01
GitHub · release · May 26

OpenAI Agents Python: v0.17.4

Release note

A new release/update from OpenAI Agents Python. Plain version: feat: support Realtime custom voice objects by @lionel-oai in * fix: #3459 add opt-in recovery for missing function tools by @seratch in * fix: apply hardened http client default to MCP SSE transport by @ioleksiuk in

Why you might care

Competitive intel: how OpenAI thinks agent SDKs should work, where Hermes is ahead/behind, and patterns we might borrow without joining the mothership.

Source excerpt

## What's Changed * feat: support Realtime custom voice objects by @lionel-oai in https://github.com/openai/openai-agents-python/pull/3473 * fix: #3459 add opt-in recovery for missing function tools by @seratch in https://github.com/openai/openai-agents-python/pull/3461 * fix: apply hardened http client default to MCP SSE transport by @ioleksiuk in https://…

agentssdkopenai/openai-agents-python
watch release58.0agent, agents, ai agent, mcp
#02
GitHub · release · Jul 1

Hermes Agent: Hermes Agent v0.18.0 (2026.7.1) — The Judgment Release

Release note

A new release/update from Hermes Agent. Plain version: Over the last week and a half the team put nearly all of its effort into one goal: resolve *…

Why you might care

Directly relevant to your agent stack. This can affect Hermes behavior, upgrade timing, plugins, gateway surfaces, or ideas worth stealing for OAC/DestructoR666.

Source excerpt

# Hermes Agent v0.18.0 (v2026.7.1) **Release Date:** July 1, 2026 **Since v0.17.0:** ~1,720 commits · 998 merged PRs · 2,215 files changed · ~251,000 insertions · ~41,000 deletions · **949 issues closed** · **370+ community contributors** > **The Judgment Release.** Over the last week and a half the team put nearly all of its effort into one goal: resolve *…

hermesagent-opsNousResearch/hermes-agent
watch release57.1hermes, hermes
#03
GitHub · release · Jun 19

Hermes Agent: Hermes Agent v0.17.0 (v2026.6.19)

Release note

A new release/update from Hermes Agent. Plain version: The Reach Release. v0.16.0 put Hermes on your desktop. v0.17.0 is about how far that reach extends — across new places to…

Why you might care

Directly relevant to your agent stack. This can affect Hermes behavior, upgrade timing, plugins, gateway surfaces, or ideas worth stealing for OAC/DestructoR666.

Source excerpt

# Hermes Agent v0.17.0 (v2026.6.19) **Release Date:** June 19, 2026 **Since v0.16.0:** ~1,475 commits · ~800 merged PRs · 1,693 files changed · 235,390 insertions · 50,730 deletions · 300+ issues closed · 245 community contributors > **The Reach Release.** v0.16.0 put Hermes on your desktop. v0.17.0 is about how far that reach extends — across new places to…

hermesagent-opsNousResearch/hermes-agent
watch release48.0hermes, hermes
#04
GitHub · release · Jul 2

Pydantic AI: v2.4.0 (2026-07-02)

Release note

A new release/update from Pydantic AI. Plain version: 🚀 Features * Split preserve_file_data into allow_uploaded_files (inbound security) and an AG-UI representation opt-in by @DouweM in * Add GEval evaluator and standard quality metric rubrics for LLMJudge by @dmontagu in * Add agen…

Why you might care

Competitive intel: how OpenAI thinks agent SDKs should work, where Hermes is ahead/behind, and patterns we might borrow without joining the mothership.

Source excerpt

## What's Changed ### 🚀 Features * Split `preserve_file_data` into `allow_uploaded_files` (inbound security) and an AG-UI representation opt-in by @DouweM in https://github.com/pydantic/pydantic-ai/pull/6232 * Add `GEval` evaluator and standard quality metric rubrics for `LLMJudge` by @dmontagu in https://github.com/pydantic/pydantic-ai/pull/5129 * Add agen…

agentspythonpydantic/pydantic-ai
watch release45.3agent, agents, eval
#05
GitHub · release · May 18

OpenAI Agents Python: v0.17.3

Release note

A new release/update from OpenAI Agents Python. Plain version: fix: keep mountpoint credentials out of sandbox commands by @seratch in * fix: unify memory optional dependency import errors by @seratch in * fix: guard None text in text_message_output and add output guardrail count to RunE…

Why you might care

Competitive intel: how OpenAI thinks agent SDKs should work, where Hermes is ahead/behind, and patterns we might borrow without joining the mothership.

Source excerpt

## What's Changed * fix: keep mountpoint credentials out of sandbox commands by @seratch in https://github.com/openai/openai-agents-python/pull/3429 * fix: unify memory optional dependency import errors by @seratch in https://github.com/openai/openai-agents-python/pull/3389 * fix: guard None text in text_message_output and add output guardrail count to RunE…

agentssdkopenai/openai-agents-python
watch release44.0agents, ai agent, memory, openai agents
#06
GitHub · release · Jul 2

MCP TypeScript SDK: @modelcontextprotocol/server-legacy@2.0.0-beta.2

Release note

A new release/update from MCP TypeScript SDK. Plain version: Patch Changes - #2405 f172626 Thanks @mattzcarey!

Why you might care

MCP is becoming the plumbing layer for agent tools. Worth watching because it shapes what Hermes/OAC can connect to without bespoke glue code.

Source excerpt

### Patch Changes - [#2405](https://github.com/modelcontextprotocol/typescript-sdk/pull/2405) [`f172626`](https://github.com/modelcontextprotocol/typescript-sdk/commit/f172626a8e98b2ae2f0f690e4afb4dc74dbf6011) Thanks [@mattzcarey](https://github.com/mattzcarey)! - Ship CommonJS builds alongside ESM. Each package now emits both `.mjs`/`.d.mts` and `.cjs`/`.d…

mcptoolsmodelcontextprotocol/typescript-sdk
skim changelog40.3mcp
#07
GitHub · release · Jul 2

MCP TypeScript SDK: @modelcontextprotocol/node@2.0.0-beta.2

Release note

A new release/update from MCP TypeScript SDK. Plain version: Patch Changes - #2405 f172626 Thanks @mattzcarey!

Why you might care

MCP is becoming the plumbing layer for agent tools. Worth watching because it shapes what Hermes/OAC can connect to without bespoke glue code.

Source excerpt

### Patch Changes - [#2405](https://github.com/modelcontextprotocol/typescript-sdk/pull/2405) [`f172626`](https://github.com/modelcontextprotocol/typescript-sdk/commit/f172626a8e98b2ae2f0f690e4afb4dc74dbf6011) Thanks [@mattzcarey](https://github.com/mattzcarey)! - Ship CommonJS builds alongside ESM. Each package now emits both `.mjs`/`.d.mts` and `.cjs`/`.d…

mcptoolsmodelcontextprotocol/typescript-sdk
skim changelog40.3mcp
#08
GitHub · release · Jun 30

MCP Python SDK: v2.0.0b1

Release note

A new release/update from MCP Python SDK. Plain version: First v2 beta, and the first release with full support for the 2026-07-28 MCP specification.

Why you might care

MCP is becoming the plumbing layer for agent tools. Worth watching because it shapes what Hermes/OAC can connect to without bespoke glue code.

Source excerpt

First v2 beta, and the first release with full support for the [2026-07-28 MCP specification](https://blog.modelcontextprotocol.io/posts/2026-07-28-release-candidate/). Pre-releases are opt-in only; `pip install mcp` still resolves to the stable 1.x line. ```bash pip install mcp==2.0.0b1 # or uv add "mcp==2.0.0b1" ``` The [documentation](https://py.sdk.mode…

mcptoolsmodelcontextprotocol/python-sdk
skim changelog37.5mcp
#09
GitHub · release · Jun 29

LangGraph: langgraph==1.2.7

Release note

A new release/update from LangGraph. Plain version: 1.2.6 * release(langgraph): 1.2.7 (#8223) * fix(langgraph): snapshot DeltaChannel overwrite supersteps (#8125) * fix(langgraph): Make Overwrite survive JSON roundtrips (#8127) * chore(deps): bump redis in /libs/langgraph (#7976)…

Why you might care

Useful as a contrast model for workflow/stateful agents. Even if we do not adopt it, it exposes design patterns and footguns for OAC-style continuity.

Source excerpt

Changes since 1.2.6 * release(langgraph): 1.2.7 (#8223) * fix(langgraph): snapshot `DeltaChannel` overwrite supersteps (#8125) * fix(langgraph): Make `Overwrite` survive JSON roundtrips (#8127) * chore(deps): bump redis in /libs/langgraph (#7976) * chore(deps): bump langsmith from 0.8.0 to 0.8.18 in /libs/langgraph (#8176) * chore(deps): bump jupyterlab fro…

agentsworkflowlangchain-ai/langgraph
skim changelog35.9agents, workflow
#10
GitHub · release · Jun 30

Pydantic AI: v2.2.0 (2026-06-30)

Release note

A new release/update from Pydantic AI. Plain version: 🚀 Features * feat: add TwelveLabs Pegasus video-understanding example by @mohit-twelvelabs in * Add Claude Sonnet 5 (claude-sonnet-5) support by @dsfaccini in * Allow factory functions as lifecycle arg for Dataset.evaluate by @nl…

Why you might care

Competitive intel: how OpenAI thinks agent SDKs should work, where Hermes is ahead/behind, and patterns we might borrow without joining the mothership.

Source excerpt

## What's Changed ### 🚀 Features * feat: add TwelveLabs Pegasus video-understanding example by @mohit-twelvelabs in https://github.com/pydantic/pydantic-ai/pull/6061 * Add Claude Sonnet 5 (`claude-sonnet-5`) support by @dsfaccini in https://github.com/pydantic/pydantic-ai/pull/6161 * Allow factory functions as lifecycle arg for Dataset.evaluate by @nlsfnr i…

agentspythonpydantic/pydantic-ai
skim changelog34.6agents, eval

Radar stats

V0 uses dumb-but-transparent scoring. Future versions can add LLM judgment after collection is bounded.
80total leads
24printed
11sources hit
$0X API spend
23feed12hn8paper8reddit29release

Editorial stance

Twitter is a goblin with binoculars, not the editorial board. V0 watches free sources first: RSS, Reddit, GitHub, HN, arXiv. Articles, research, opinionated field notes, and real user pain reports get top billing; releases are parked lower because staying current matters, but changelog spelunking is not a personality.

Collector logs

Receipts and failures. The web remains held together by raccoon string.