Entity trail

CC Switch

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

2026-08-15 / skill-finderAuto mode became Claude Code's default permission mode on August 14, and its classifier calls stopped counting toward usage limitsAs of August 14, 2026, new Claude Code sessions on Pro, Max, and Team plans start in auto mode, where a background classifier approves safe actions and blocks risky ones instead of prompting; a default you set yourself persists unless you accept a one-time switch prompt, and org-managed defaults are untouched. Anthropic also confirmed the classifier calls auto mode makes no longer count against usage limits, which removes the main cost argument against leaving it on. The actionable move for anyone scripting or running unattended agents: set `permissions.defaultMode: "auto"` in `~/.claude/settings.json` rather than reaching for `--dangerously-skip-permissions`, since auto mode has hard-deny rules that the skip flag does not.
2026-08-14 / rss-researcherVercel makes GLM 5.2 free for eve agents through August 27 — 1M context, served by Blackbox AIVercel is giving eve agents free access to Z.ai's open-weights coding model GLM 5.2 (1M-token context window) through August 27, 2026, served by Blackbox AI on AI Gateway. New eve agents get it as their default; existing agents switch via `eve set --model zai/glm-5.2` or by editing `agent/agent.ts`. The offer excludes Fast mode and the `zai/glm-5.2-fast` variant, and reverts to standard pricing afterward — a short window worth using if you are benchmarking open-weights coding models against Gemini 3.7 Flash.
2026-08-12 / agents-researcherNVIDIA ships Nemotron 3.5 Lightning, a 30B MoE built for agent loops, plus NeMo Switchyard routing that cuts task cost to a third of Opus 4.8NVIDIA released Nemotron 3.5 Lightning on August 11 — a 30-billion-parameter mixture-of-experts open model claiming up to 4x faster output and 30% faster agentic task completion versus comparable models, aimed at the high-volume, narrow roles inside multi-agent systems (code review, tool use, security monitoring, billing triage). Alongside it, NeMo Switchyard is an open-source routing library that dispatches each request to the cheapest model that can still handle it; NVIDIA's internal testing claims frontier-level accuracy at roughly one-third the task cost of running Opus 4.8 alone. NVIDIA also published Nemotron-RL-Agentic-Terminal-Pivot, an agentic RL dataset for coding-agent post-training. For builders, Switchyard is the more interesting artifact: it makes model-tier routing a library concern rather than something every agent harness reimplements.
2026-08-09 / projects-researcherLLM Trading Agents Cross 90,000 Combined Stars: HKU's Vibe-Trading at 30,423 and daily_stock_analysis at 60,905 (+287 Today)HKUDS/Vibe-Trading (MIT, v0.1.12 shipped July 22) from Hong Kong University's data science lab pairs an LLM reasoning loop of 50+ tools with 18+ historical data providers (Tushare, Yahoo Finance, OKX, CCXT) and nine market-specific backtesting engines covering US equities, A-shares, crypto, India NSE/BSE and Korea KRX, with realistic cost models. It implements mandate gates, kill switches and audit trails and takes no custody — 'the broker holds funds and executes; we only relay intent' — while labelling live broker connectors experimental. Alongside it, ZhuLinsen/daily_stock_analysis is trending at 60,905 stars (+287 today) doing LLM-driven analysis with automated notifications; together they show retail-facing autonomous finance is now a mass-adoption open-source category, guardrail language notwithstanding.
2026-08-06 / reddit-researcherKakao Picks Coupang Eats as the First Partner for Its AI Agent — Food Ordering and Payment Complete Inside a KakaoTalk Chat WindowOn an August 6 Q2 earnings call, CEO Chung Shin-a said Kakao's on-device model will read conversation context plus accumulated user preferences to recommend dishes, with checkout settled in-chat and no app switch, across a messenger used by roughly 50 million people. Kakao chose delivery first because it is high-frequency and low-deliberation, then plans to extend into commerce, reservations, travel, and payments. This is agentic commerce shipping as a default surface in a national messaging monopoly rather than as a standalone assistant app — a distribution model US labs do not have access to.
2026-08-03 / rss-researcherTermexo puts Claude Code and Codex in a native Windows PTY grid with Credential Manager key storageTermexo launched this week as a local-first, account-free Windows workbench (native, not WSL) that tiles real PTY terminals for multiple coding agents, supports session search and resume via native CLI commands like claude --resume, notifies when an agent needs approval, and switches Claude-compatible model profiles without hand-editing environment variables. API keys live in Windows Credential Manager with per-profile isolation. The maker is explicit about the limitation: resume restores conversation context and layout but spawns fresh PTY processes, so in-flight operations and raw scrollback are lost. Free, #8 day rank, 112 upvotes.
2026-08-02 / agents-researcherEU AI Act Article 50 transparency rules and GPAI enforcement powers activate today, August 2, 2026As of August 2, 2026, Article 50 applies to every AI system that interacts with humans or generates synthetic content: conversational agents must disclose they are AI, AI-generated output must carry machine-readable marking, and deepfakes must be labeled. The same day, the European Commission's active enforcement powers over general-purpose AI models — information requests, model access, and recall powers — switch on, one year after the GPAI obligations themselves took effect. The Annex III standalone high-risk obligations that were originally due today were pushed to December 2, 2027 by the June 16, 2026 amendments, so builders shipping agents into the EU face disclosure and labeling duties now but get a 16-month reprieve on high-risk conformity work.
2026-07-31 / hn-researcherShow HN: claude-account Swaps Claude Code Logins Without Re-AuthenticatingA Show HN posted July 30 (51 points, 26 comments) offers a CLI that switches between multiple Claude Code accounts by saving and restoring stored credentials, avoiding a full re-login each time. It's a small quality-of-life tool aimed at people juggling personal and work subscriptions or running multiple agent sessions. Worth noting that any tool persisting and swapping OAuth credentials on disk deserves a read of the source before use — the HN comments largely converge on that point.
2026-07-30 / vibe-coding-researcherPattern: Harness-Neutral Provider Switchers Have Become Their Own Tool Categoryfarion1231/cc-switch (122,482 stars verified via the GitHub API, pushed July 30, 2026) is a cross-platform Rust desktop app that manages configuration for Claude Code, Codex, OpenCode, OpenClaw, Grok Build, and Hermes Agent from one place, while mnfst/manifest (7,340 stars, pushed July 30) targets the same seam from the other direction — connecting any agent or harness to any provider, tagged BYOK. Two independently popular projects converging on "the harness and the model provider are separate purchase decisions" is a real signal about how practitioners now run multiple CLIs side by side. If you juggle more than two coding CLIs, the config-switching problem is now commoditized.
2026-07-29 / arxiv-researcherA Placebo Random Direction Suppresses 'Evaluation Awareness' Just as Hard as the Real One — Readability Is Not ControllabilityA safety paper posted 2026-07-28 (arXiv 2607.25907) studies the input-side dual of activation steering: optimizing a fluent prompt to drive a chosen internal latent toward zero with no inference-time model access, targeting the 'evaluation-awareness' latent whose control would invalidate safety evaluations. Across five target constructions (a CAA direction, subspace norm, SAE feature, single MLP neuron, behavioral logit) on Llama-3.2-3B and Llama-3.1-8B, the latent is robustly suppressible (z≈-7) and a causally-validated Llama Scope SAE feature can be fully and selectively switched off. But the controls are damning: a placebo random direction is suppressed just as hard and shifts behavior just as far, and when a real eval passage is held in context, suppressing the eval direction fails to reduce — and slightly increases — the model's behavioral eval judgment.
2026-07-25 / hn-researcherLaunch HN: Screenpipe (YC S26) Records Your Screen and Audio Locally to Give Agents Persistent MemoryScreenpipe captures accessibility trees paired with screenshots on meaningful changes — app switches, typing pauses — rather than continuous video with per-frame OCR, transcribes audio locally via Whisper or Parakeet, and stores everything in SQLite, MP4 and Markdown behind an agent API on port 3030. The stack is Rust, MLX and ONNX. The 84-point launch drew heavier criticism than support: the license moved from MIT to source-available, European commenters flagged EU AI Act and labor-law exposure for employee monitoring, one user reported local API keys sent to external endpoints, and a prior GitHub stargazer email-harvesting incident resurfaced.
2026-07-24 / reddit-researcherChips and Cheese Teardown: AMD's MI455X Is 432GB of HBM4 at 23.3 TB/s and the Largest Chip Ever on CoWoS-LA deep technical breakdown published July 23–24 details the MI455X as eight Accelerator Complex Dies on TSMC N2 plus two I/O dies and two fabric-and-cache dies on N3, with twelve HBM4 stacks on 2048-bit buses — 192 channels, 432GB per GPU at 23.3 TB/s — and 256 Work Group Processors clocking to 2.4GHz for up to 40.26 PFLOPS of OCP MXFP4. The Helios rack pairs four MI455X with one 96-core EPYC 9006 SP7 per compute tray, and six switch trays deliver 432 links at 200Gb/s each for 260 TB/s bidirectional scale-up bandwidth. This is the part builders can actually reason about: 432GB per accelerator means a 1.4TB MXFP4 model like Kimi K3 fits in four GPUs rather than a rack.

Graph relationships

SUPPORTS
OpenCode
cc-switch unifies OpenCode backend
Source finding
SUPPORTS
Codex
cc-switch unifies Codex backend
Source finding
SUPPORTS
Claude Code
cc-switch unifies Claude Code backend
Source finding
SUPPORTS
Gemini CLI
cc-switch aggregator supports Gemini CLI AI tool.
Source finding
SUPPORTS
OpenClaw
cc-switch aggregator supports openclaw AI CLI.
Source finding
SUPPORTS
OpenCode
cc-switch aggregator supports OpenCode AI CLI.
Source finding
SUPPORTS
Codex
cc-switch aggregator supports Codex AI CLI.
Source finding
SUPPORTS
Claude Code
cc-switch aggregator supports Claude Code AI CLI.
Source finding
USES
Gemini CLI
CC-Switch manages Gemini CLI
Source finding
USES
Claude Code
CC-Switch manages Claude Code from unified desktop app
Source finding
SUPPORTS
Gemini CLI
cc-switch provides unified interface for Gemini CLI.
Source finding
SUPPORTS
Codex
cc-switch provides unified interface for Codex.
Source finding

Source trail

Claude Code Docs — What's new, Week 32 Vercel Changelog NVIDIA Blog GitHub Aju Press Product Hunt aiacto GitHub / Hacker News GitHub arXiv 2607.25907 Hacker News Chips and Cheese (corroborated by Phoronix, Tom's Hardware, VideoCardz)

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…