Heads up This site is currently under heavy development.
Subscribe Get it delivered — the daily firehose, filtered to the tools you run, plus the documentation changes vendors never announce. Compare plans →

The AI Toolchain — issue -028, July 22, 2026

THE AI TOOLCHAIN NO. -028
Tail
THE DAILY RELEASE FIREHOSE
PUBLISHED JULY 22, 2026 · EVERY WEEKDAY
EDITIONS tail grep head diff uniq

The daily firehose — everything the toolchain shipped today, already filtered.

// HOW THIS ISSUE IS MADE

We read every release from the 174 tools on our watchlist at the source — GitHub and GitLab release notes, vendor release pages and changelogs, project blogs and feeds, vendor press releases, and the source code behind the tag. Bug-fix-only releases and non-product newsroom noise are dropped; what's left is summarized down to the new capability, how to try it, and any screenshots or videos the release itself published. Every entry links to the sources it was built from.

VIEW
ISSUE VIEW full issue
Do you prefer this view?
$ tct list   # 8 tools matched
AI & LLM Tooling
◆  AI Agent Frameworks

AutoGPT

Sources Release notes → autogpt-platform-beta-v0.6.69 NOTES

AutoGPT Platform adds proactive Slack/Telegram posting, Agent-Building Mode, named batch deploys, and GPT-5 model support.

└──▷ GET THIS VERSION
$ git clone --branch autogpt-platform-beta-v0.6.69 https://github.com/Significant-Gravitas/AutoGPT.git
# already have the repo? check out this version:
$ git checkout autogpt-platform-beta-v0.6.69
  • Adds support for GPT-5.6, GPT-5.5, GPT-5.4, and o-series OpenAI models as selectable options.
  • Introduces multi-batch support for the batch-deploy bot with named batches, capped at 4 concurrent batches.
  • Enables proactive posting to Slack and Telegram channels from the copilot bot, including DM delivery for proactive posts.
  • Adds Agent-Building Mode with a compaction-proof guide and automatic engine switching.
  • Streamlines AutoPilot agent creation flow.
+2 moreshow less
  • Adds a redesigned app sidebar layout with a new profile menu, org switcher, and agent activity panel.
  • Displays block name instead of block ID during run_block in the UI.
Was this useful?

PydanticAI

Sources Release notes → v2.15.0 NOTES

PydanticAI v2.15.0 adds per-run tool-retry budget overrides, OpenAI moderation, and DynamicCapability toolset support in durable execution.

└──▷ GET THIS VERSION
$ git clone --branch v2.15.0 https://github.com/pydantic/pydantic-ai.git
# already have the repo? check out this version:
$ git checkout v2.15.0
└──▷ TRY IT
Cap tool retries for a single high-stakes run without changing the agent's default retry budget.
$ result = await agent.run('Fetch and summarize the report', tool_retry_budget=2)
Enable OpenAI Responses API moderation and inspect the result on the model response.
python
result = await agent.run('Draft a message', model_settings={'openai_moderation': True})
print(result.all_messages()[-1].provider_details)
  • Adds openai_moderation setting to expose OpenAI Responses API moderation results in provider_details.
  • Supports overriding the tool-retry budget at run, iter, and override time, giving per-invocation control over retry limits.
  • Supports DynamicCapability toolsets in durable execution, wrapping DynamicToolset in DBOS steps and Prefect tasks.
  • Adds explicit prompt caching support for gpt-5.6 in OpenAIModel.
  • Inlines text-like files in MistralModel prompts for cleaner multimodal input handling.
+1 moreshow less
  • Introduces ExaSearch capability in Pydantic AI Harness as the successor to the Exa search common tools.
Was this useful?
◆  AI Coding Agents

Anthropic Claude Code

Sources Release notes → v2.1.218 NOTES

Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflows - all through natural language commands.

Claude Code v2.1.218 runs /code-review as a background subagent and expands skill/plugin config with new boolean values

└──▷ GET THIS VERSION
$ git clone --branch v2.1.218 https://github.com/anthropics/claude-code.git
# already have the repo? check out this version:
$ git checkout v2.1.218
└──▷ USE IT
Opt a specific skill out of the new default background execution when it uses context: fork.
yaml
context: fork
background: false
Diagnose a failing MCP server connection — HTTP status and error text now appear inline.
$ claude mcp list
  • Accepts yes/no/on/off/1/0 (case-insensitive) as valid boolean values for skill and plugin frontmatter, alongside true/false.
  • Skills with context: fork now run in the background by default; opt out per skill with background: false in frontmatter.
  • Agent markdown files now reject agent names containing : (reserved for plugin namespacing).
  • Adds HTTP status and error text to claude mcp list and /mcp output when a server fails to connect, plus a warning for MCP config values with hidden leading or trailing whitespace.
  • Changes /code-review to run as a background subagent, keeping stacked slash commands as its review target without filling the active conversation.
+7 moreshow less
  • Changes /deep-research to start only on manual invocation; Claude no longer auto-launches it.
  • Adds screen-reader announcements of deleted text for word and line deletions (Option+Delete, Ctrl+W, Cmd+Backspace, Ctrl+U, Ctrl+K) in --ax-screen-reader mode.
  • Adds an announcement when fast mode changes as a result of switching models via /config model=<x> or Remote Control.
  • Changes plan mode with auto so it no longer prompts for Bash commands the static analyzer cannot prove read-only; the auto-mode classifier judges them instead.
  • Improves auto mode: dangerous-rm, background-&, and suspicious-Windows-path checks no longer open permission dialogs — the auto-mode classifier adjudicates them.
  • Improves trust dialogs to name the repository root the grant covers.
  • Changes server-managed settings so benign feature and cost toggles no longer trigger the settings-approval prompt.
└──▷ BREAKING ON UPGRADE
  • !Agent markdown files that use : in agent names will now be rejected; rename any such agents before upgrading.
Was this useful?

Cline

Sources Release notes → desktop-v0.0.2 NOTES

Autonomous coding agent as an SDK, IDE extension, or CLI assistant.

Cline Code desktop app launches for macOS with automatic background updates for Apple Silicon and Intel.

└──▷ GET THIS VERSION
$ git clone --branch desktop-v0.0.2 https://github.com/cline/cline.git
# already have the repo? check out this version:
$ git checkout desktop-v0.0.2
  • First public release of Cline Code for macOS: a signed and notarized desktop app for running and inspecting Cline agent sessions, available as a DMG for Apple Silicon and Intel.
  • Automatic update delivery: the app checks for new versions on launch and every 2 hours, downloads them in the background, and prompts for a one-click restart; ignored updates apply on the next launch.
Was this useful?

OpenAI Codex CLI

Sources Release notes →Source code → rust-v0.145.0-alpha.30 NOTES CODE PRE-RELEASE

Lightweight coding agent that runs in your terminal

Codex CLI alpha.30 adds Esc-key session interruption and parallel stdin writes across unified-exec sessions.

└──▷ GET THIS VERSION
$ git clone --branch rust-v0.145.0-alpha.30 https://github.com/openai/codex.git
# already have the repo? check out this version:
$ git checkout rust-v0.145.0-alpha.30
  • Adds Esc key as an interrupt mechanism to cancel a running agent task mid-session, now documented in the status indicator.
  • Runs write_stdin in parallel across unified-exec sessions, improving throughput when multiple shell sessions are active.
  • Updates bundled GPT-5.6 model variants with 272,000-token context windows and adds personality instruction variables.
  • Adds GPT-5.5 availability notice to the bundled model catalog.
└──▷ BREAKING ON UPGRADE
  • !The auto_review and permissions message fields and the legacy supports_reasoning_summary_parameter flag are removed from the bundled model catalog.
Was this useful?
◆  Local LLM Runtimes

llama.cpp

Sources Release notes → b10090 3 RELEASES · 2026-07-22 NOTES STABLE

llama.cpp b10090 adds GGML_OP_CONV_2D_DW (depthwise conv2d) support to the WebGPU backend.

└──▷ GET THIS VERSION
$ git clone --branch b10090 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b10090
  • Implements GGML_OP_CONV_2D_DW (depthwise convolution 2D) kernel for the WebGPU backend, ported from the Vulkan backend's conv2d_dw.comp shader.
2 more releases in this issue · 2026-07-22
b10089 NOTES STABLE

CUDA GET_ROWS now handles all quantized GGML types on-device, eliminating host fallback for k-quants, i-quants, and mxfp4.

└──▷ GET THIS VERSION
$ git clone --branch b10089 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b10089
  • Adds full k-quant support (q2_K through q6_K) to the CUDA GET_ROWS kernel via a new k_get_rows_kq kernel, keeping embedding lookups on-device instead of falling back to the host and copying the full embedding matrix per token.
  • Extends CUDA GET_ROWS device-side coverage to all nine i-quants, completing quantized type support alongside k-quants.
  • Adds mxfp4 support to CUDA GET_ROWS, closing the final gap so every quantized GGML type now takes the direct device path in single-device graphs.
b10087 NOTES STABLE

llama.cpp b10087 adds support for Laguna XS.2 & M.1 hardware.

└──▷ GET THIS VERSION
$ git clone --branch b10087 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b10087
  • Adds support for Laguna XS.2 & M.1 accelerator hardware.
Was this useful?
Other / Uncategorized
◆  AI OBSERVABILITY

Arize Phoenix

Sources Release notes → arize-phoenix-v19.4.0 NOTES

Phoenix v19.4.0 upgrades JS tracing to AI SDK v7 and redesigns the span header UI.

└──▷ GET THIS VERSION
$ git clone --branch arize-phoenix-v19.4.0 https://github.com/Arize-ai/phoenix.git
# already have the repo? check out this version:
$ git checkout arize-phoenix-v19.4.0
  • Upgrades phoenix-otel to openinference-vercel v3, adding support for AI SDK v7 spans in JavaScript tracing.
  • Redesigns the span header into separate identity and meta rows in the tracing UI.
  • Makes IDBadge the consolidated click-to-copy ID element across the UI.
  • Aligns span kind icon and timeline bar colors with token treatment in the components UI.
Was this useful?

Langfuse

Sources Release notes → v3.224.0 NOTES

Langfuse v3.224.0 adds boolean score widgets, OTEL media upload, streaming trace I/O, and a mobile shell.

└──▷ GET THIS VERSION
$ git clone --branch v3.224.0 https://github.com/langfuse/langfuse.git
# already have the repo? check out this version:
$ git checkout v3.224.0
  • Streams trace observation input/output as bytes via the API (feat(api): stream trace observation IO as bytes).
  • Uploads media during OTEL worker ingestion, enabling media attachment through the OpenTelemetry pipeline.
  • Adds boolean score widgets and monitors to dashboards.
  • Migrates MCP score read tools from scores v2 to v3.
  • Reads prompt usage counts from events using the v4 prompt pipeline.
+1 moreshow less
  • Introduces a minimal-chrome mobile shell for improved mobile layout.
Was this useful?
my-toolchain — 0 tools
paste an install list to detect your tools

A brew list, a Brewfile, requirements.txt, a Dockerfile — or just the product names, free-form. Nothing leaves your browser.

    browse all tools →