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

Signal
Rag

Jun 18, 2026 80 leads scanned 8 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 · Jun 18

Show HN: Automatiq - generate webscrapers/automations by browsing any website

Editor's read

A community-submitted project/thread. Plain version: A tool that watches you browse, then writes HTTP-based automation scripts - StoneSteel27/AutomatiQ

Why you might care

HN is noisy, but comments surface implementation pain fast. Good for spotting whether a shiny thing survives contact with engineers.

Source excerpt

A tool that watches you browse, then writes HTTP-based automation scripts - StoneSteel27/AutomatiQ

hnAI agents tool usedeep-summary
read44.6agent, agents, tool use
#02
Hacker News · hn · Jun 17

Show HN: We cut >60% of tokens from agentic tasks by removing repeated context

Editor's read

A community-submitted project/thread. Plain version: Connect your databases, documents, Google Drive, Notion etc.

Why you might care

Directly relevant to context bloat and tool-loading discipline. If it is good, it may inform how Hermes/OAC keeps prompts lean.

Source excerpt

Connect your databases, documents, Google Drive, Notion etc. so your AI systems pull the right data, stop hallucinating, and deliver results you can actually trust.

hnAI agents tool usedeep-summary
read44.4agent, agents, tool use
#03
Hacker News · hn · Jun 4

Bad MCP design costs your agent 5x more tokens

Editor's read

A community-submitted project/thread. A source lead with no useful publisher summary. Use the title as the hook, then only open it if the topic matches today’s agent/tooling work: Bad MCP design costs your agent 5x more tokens. Treat it as an early smoke signal, not a verdict. HN: 17 points, 1 comments. Query: LLM eval benchmark.

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: 17 points, 1 comments. Query: LLM eval benchmark.

hnLLM eval benchmark
read44.0agent, mcp, eval, benchmark
#04
arXiv · paper · Jun 17

UBP2: Uncertainty-Balanced Preference Planning for Efficient Preference-based Reinforcement Learning

Editor's read

A research paper. In human terms: Preference-based RL provides an approach to learning reward models from pairwise comparisons of behaviors, bypassing the need for explicit reward design.

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

Preference-based RL provides an approach to learning reward models from pairwise comparisons of behaviors, bypassing the need for explicit reward design. However, existing methods typically rely on passive data collection and suffer from poor sample efficiency, especially during the early stages of learning. We introduce a model-based approach that actively directs exploration by jointly reasoning over uncertainties…

paperLLM evaluation benchmark agentscs.LGcs.AI
paper shelf38.6agent, agents, eval, benchmark
#05
arXiv · paper · Jun 17

Rethinking Reward Supervision: Rubric-Conditioned Self-Distillation

Editor's read

A research paper. In human terms: Post-training of reasoning language models is commonly driven by supervised distillation and reinforcement learning with verifiable rewards.

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

Post-training of reasoning language models is commonly driven by supervised distillation and reinforcement learning with verifiable rewards. Distillation often relies on chain-of-thought annotations that are expensive to obtain and may themselves be noisy, incomplete, or partially incorrect; even when the final solution is correct, an imperfect rationale can interfere with learning. Reinforcement learning with verif…

paperLLM evaluation benchmark agentscs.AIcs.CL
paper shelf38.6agent, agents, eval, benchmark

Articles / research / opinions

The stuff people are thinking, trying, arguing, and building with the tech. This is the main read pile.
#06
arXiv · paper · Jun 17

Zero-Shot Long-Horizon Dexterous Manipulation via Multi-View 3D-Grounded VLM Reasoning

Editor's read

A research paper. In human terms: We present a zero-shot framework for long-horizon dexterous manipulation that grounds language instructions into executable 3D task plans from calibrated multi-view RGB images.

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 a zero-shot framework for long-horizon dexterous manipulation that grounds language instructions into executable 3D task plans from calibrated multi-view RGB images. Rather than training an end-to-end policy, our system uses a vision-language model (VLM) to produce reference-frame task grounding and primitive-level 2D keypoints, then lifts them into 3D via multi-view fusion. This lifting combines triangul…

paperlarge language model agents memorycs.RO
paper shelf34.6agent, agents, memory
#07
Simon Willison · feed · Jun 15

Cloudflare CAPTCHA on at least one ampersand

Editor's read

A post from Simon Willison. Plain version: TIL: Cloudflare CAPTCHA on at least one ampersand I'm using Cloudflare's CAPTCHA (they call it a "Web Application Firewall > Custom rules > Managed Challenge" these days) to prevent crawlers from aggresively s…

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

TIL: Cloudflare CAPTCHA on at least one ampersand I'm using Cloudflare's CAPTCHA (they call it a "Web Application Firewall > Custom rules > Managed Challenge" these days) to prevent crawlers from aggresively spidering my faceted search engine on this site, but I got fed up of even simple ?q=term searches triggering the challenge. After some mucking around w…

llmtoolspractical
skim34.5feed
#08
OpenAI News · feed · Jun 10

BBVA puts AI at the core of banking with OpenAI

Editor's read

A post from OpenAI News. Plain version: Learn how BBVA scaled ChatGPT Enterprise to 100,000 employees and partnered with OpenAI to accelerate AI-powered banking transformation worldwide.

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

Learn how BBVA scaled ChatGPT Enterprise to 100,000 employees and partnered with OpenAI to accelerate AI-powered banking transformation worldwide.

openaimodelsagents
skim34.0agents
#09
OpenAI News · feed · Jun 10

Supporting Europe’s work in ensuring a trustworthy AI ecosystem

Editor's read

A post from OpenAI News. Plain version: OpenAI supports the EU Code of Practice on AI content transparency, advancing provenance standards and tools to help people understand AI-generated content.

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 supports the EU Code of Practice on AI content transparency, advancing provenance standards and tools to help people understand AI-generated content.

openaimodelsagents
skim34.0agents
#10
OpenAI News · feed · Jun 10

PRC-linked influence operations are targeting AI debates in the US

Editor's read

A post from OpenAI News. Plain version: A new report from OpenAI details PRC-linked influence operations using AI to target U.S.

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

A new report from OpenAI details PRC-linked influence operations using AI to target U.S. tech debates, data center narratives, tariffs, and false claims about ChatGPT.

openaimodelsagents
skim34.0agents
#11
Latent Space · feed · Jun 9

[AINews] Anthropic Claude Fable 5 — Mythos but Safe, with Controversial Terms

Editor's read

A post from Latent Space. Plain version: The much anticipated launch of the Mythos-class model was marred by some controversial usage policies

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 much anticipated launch of the Mythos-class model was marred by some controversial usage policies

aiindustryagents
skim34.0agents
#12
Hugging Face Blog · feed · Jun 9

How an Agent Built a 3D Paris Gallery by Chaining Two Hugging Face Spaces

Editor's read

A post from Hugging Face Blog. Plain version: A source lead with no useful publisher summary. Use the title as the hook, then only open it if the topic matches today’s agent/tooling work: How an Agent Built a 3D Paris Gallery by Chaining Two Hugging Face Spaces.

Why you might care

Potentially useful for keeping your stack cheaper, more private, and less dependent on cloud APIs. Worth a skim if it looks practical.

Source excerpt

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

modelsopen-sourceinference
skim34.0inference
#13
Hugging Face Blog · feed · Jun 8

Migrating Your GitHub CI to Hugging Face Jobs

Editor's read

A post from Hugging Face Blog. Plain version: A source lead with no useful publisher summary. Use the title as the hook, then only open it if the topic matches today’s agent/tooling work: Migrating Your GitHub CI to Hugging Face Jobs.

Why you might care

Potentially useful for keeping your stack cheaper, more private, and less dependent on cloud APIs. Worth a skim if it looks practical.

Source excerpt

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

modelsopen-sourceinference
skim34.0inference
#14
Latent Space · feed · Jun 5

[AINews] not much happened today

Editor's read

A post from Latent Space. Plain version: A source lead with no useful publisher summary. Use the title as the hook, then only open it if the topic matches today’s agent/tooling work: [AINews] not much happened today.

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

a quiet day of RSI.

aiindustryagents
skim34.0agents

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 release51.0agent, agents, mcp, openai agents
#02
GitHub · release · Jun 5

Hermes Agent: Hermes Agent v0.16.0 (2026.6.5) — The Surface Release

Release note

A new release/update from Hermes Agent. Plain version: The Surface Release. Hermes meets you wherever you work. A brand-new na…

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 v0.16.0 (v2026.6.5) **Release Date:** June 5, 2026 **Since v0.15.2:** 874 commits · 542 merged PRs · 1,962 files changed · 205,216 insertions · 46,217 deletions · 399 issues closed (2 P0, 62 P1, 16 security-tagged) · 170 community contributors (including co-authors) > **The Surface Release.** Hermes meets you wherever you work. A brand-new na…

hermesagent-opsNousResearch/hermes-agent
watch release48.0hermes, hermes
#03
GitHub · release · May 29

Hermes Agent: Hermes Agent v0.15.2 (2026.5.29.2)

Release note

A new release/update from Hermes Agent. Plain version: Packaging): ship bundled plugin.yaml manifests in wheel and sdist (827f7f07) ## 👥 Contributors Thank you to everyone who contributed to this release!

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 v0.15.2 (v2026.5.29.2) **Release Date:** May 29, 2026 ## 🐛 Bug Fixes - Packaging): ship bundled plugin.yaml manifests in wheel and sdist ([`827f7f07`](https://github.com/NousResearch/hermes-agent/commit/827f7f07825be57108cbea18325e8f5e9fb5d2f2)) ## 👥 Contributors Thank you to everyone who contributed to this release! - @outsourc-e (1 commit)…

hermesagent-opsNousResearch/hermes-agent
watch release48.0hermes, hermes
#04
GitHub · release · Jun 16

MCP Python SDK: v2.0.0a2

Release note

A new release/update from MCP Python SDK. Plain version: Full 2026-07-28 types added…

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

Second v2 alpha. Pre-releases are opt-in only; `pip install mcp` still resolves to the stable 1.x line. ```bash pip install mcp==2.0.0a2 # or uv add "mcp==2.0.0a2" ``` See the [migration guide](https://github.com/modelcontextprotocol/python-sdk/blob/main/docs/migration.md) for the full list of breaking changes. ## Highlights ### Full 2026-07-28 types added…

mcptoolsmodelcontextprotocol/python-sdk
skim changelog39.1mcp
#05
GitHub · release · Jun 16

MCP Python SDK: v1.28.0

Release note

A new release/update from MCP Python SDK. Plain version: Deprecations Two API surfaces now emit DeprecationWarning ahead of their removal in v2.

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

## Deprecations Two API surfaces now emit `DeprecationWarning` ahead of their removal in v2. Nothing is removed in 1.x, and the warnings fire only when the deprecated API is *called* - importing the modules stays silent. - **WebSocket transport** - `mcp.client.websocket.websocket_client` and `mcp.server.websocket.websocket_server`. WebSocket was never part…

mcptoolsmodelcontextprotocol/python-sdk
skim changelog39.1mcp
#06
GitHub · release · Jun 16

LangGraph: langgraph-cli==0.4.30

Release note

A new release/update from LangGraph. Plain version: cli==0.4.29 * release(cli): 0.4.30 (#8101) * feat(cli): support compatible api version ranges (#8023) * docs: standardize package README.md structure (#8064)

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 cli==0.4.29 * release(cli): 0.4.30 (#8101) * feat(cli): support compatible api version ranges (#8023) * docs: standardize package `README.md` structure (#8064)

agentsworkflowlangchain-ai/langgraph
skim changelog39.0agents, workflow
#07
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
skim changelog37.0agents, memory, openai agents
#08
GitHub · release · Apr 1

MCP TypeScript SDK: @modelcontextprotocol/node@2.0.0-alpha.1

Release note

A new release/update from MCP TypeScript SDK. Plain version: Patch Changes - #1504 327243c Thanks @corvid-agent!

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 - [#1504](https://github.com/modelcontextprotocol/typescript-sdk/pull/1504) [`327243c`](https://github.com/modelcontextprotocol/typescript-sdk/commit/327243cebd96e07686c88f7fa9ca22a5a7a7993d) Thanks [@corvid-agent](https://github.com/corvid-agent)! - Add missing `hono` peer dependency to `@modelcontextprotocol/node`. The package already de…

mcptoolsmodelcontextprotocol/typescript-sdk
skim changelog37.0agent, mcp
#09
GitHub · release · Jun 12

LangGraph: langgraph==1.2.5

Release note

A new release/update from LangGraph. Plain version: 1.2.4 * release(langgraph): 1.2.5 (#8062) * fix(langgraph): merge lc_versions config metadata (#8052) * release(cli): 0.4.28 (#8041) * fix: updateState bug for deltaChannel on empty thread (#8011) * chore: migrate Python type che…

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 1.2.4 * release(langgraph): 1.2.5 (#8062) * fix(langgraph): merge `lc_versions` config metadata (#8052) * release(cli): 0.4.28 (#8041) * fix: updateState bug for deltaChannel on empty thread (#8011) * chore: migrate Python type checking to ty (#8002) * chore(deps-dev): bump types-requests from 2.33.0.20260408 to 2.33.0.20260518 in /libs/langgr…

agentsworkflowlangchain-ai/langgraph
skim changelog31.8agents, workflow
#10
GitHub · release · Apr 1

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

Release note

A new release/update from MCP TypeScript SDK. Plain version: Patch Changes - #1840 424cbae Thanks @KKonstantinov!

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 - [#1840](https://github.com/modelcontextprotocol/typescript-sdk/pull/1840) [`424cbae`](https://github.com/modelcontextprotocol/typescript-sdk/commit/424cbaeee13b7fe18d38048295135395b9ad81bb) Thanks [@KKonstantinov](https://github.com/KKonstantinov)! - tsdown exports resolution fix

mcptoolsmodelcontextprotocol/typescript-sdk
skim changelog30.0mcp

Radar stats

V0 uses dumb-but-transparent scoring. Future versions can add LLM judgment after collection is bounded.
80total leads
24printed
8sources hit
$0X API spend
23feed23hn4paper30release

Editorial stance

Twitter is a goblin with binoculars, not the editorial board. V0 watches free sources first: RSS, GitHub, HN, arXiv. Articles, research, and opinionated field notes 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.