The AI Toolchain — issue -147, March 25, 2026

AutoGPT Platform adds workflow import from n8n/Make/Zapier, dry-run mode, parallel execution, and admin marketplace controls.

$ git clone --branch autogpt-platform-beta-v0.6.53 https://github.com/Significant-Gravitas/AutoGPT.git
# already have the repo? check out this version:
$ git checkout autogpt-platform-beta-v0.6.53

• ›Adds dry-run execution mode with LLM block simulation, letting teams validate agent logic without live API calls.
• ›Adds workflow import from external automation tools (n8n, Make.com, Zapier), enabling migration of existing automations into AutoGPT Platform.
• ›Adds tool/block capability filtering to AutoPilotBlock, giving operators fine-grained control over which blocks an autopilot session can invoke.
• ›Enables parallel block execution via infrastructure-level pre-launch, unlocking concurrent step processing in agent graphs.
• ›Adds /pr-test skill with subscription mode auto-provisioning for automated pull-request validation workflows.

langchain-openrouter 0.2.0 adds app_categories field for marketplace attribution and new model-profile fields.

$ git clone --branch langchain-openrouter==0.2.0 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:
$ git checkout langchain-openrouter==0.2.0

• ›Adds app_categories field to the OpenRouter integration for marketplace attribution.
• ›Adds new fields to model profiles.

LlamaIndex v0.14.19 adds MiniMax LLM, Azure OpenAI Responses API support, LiteLLM custom providers, and GPT-5.4 Mini/Nano variants.

$ git clone --branch v0.14.19 https://github.com/run-llama/llama_index.git
# already have the repo? check out this version:
$ git checkout v0.14.19

from llama_index.llms.minimax import MiniMax

llm = MiniMax()  # defaults to M2.7
response = llm.complete('Summarize the OWASP Top 10 for 2025.')
print(response)

from llama_index.llms.litellm import LiteLLM

llm = LiteLLM(
    model='openai/gpt-4o',
    model_kwargs={'custom_llm_provider': 'azure'}
)
response = llm.complete('List the top cloud misconfigurations.')
print(response)

• ›Adds llama-index-llms-minimax integration (v0.1.0) with MiniMax LLM provider, defaulting to the M2.7 model.
• ›Adds Azure OpenAI Responses API support in llama-index-llms-azure-openai.
• ›Adds support for a custom LLM provider via model kwargs in llama-index-llms-litellm.
• ›Adds support for Mini and Nano variants of GPT-5.4 in llama-index-llms-openai.
• ›Updates llama-index-llms-google-genai to default to Gemini 3 and exposes temperature control.

OpenAI Agents SDK v0.13.1 adds an any-llm adapter to the extension module for responses-compatible multi-LLM routing.

$ git clone --branch v0.13.1 https://github.com/openai/openai-agents-python.git
# already have the repo? check out this version:
$ git checkout v0.13.1

• ›Adds an any-llm adapter to the extension module, enabling responses-compatible routing to any LLM supported by the [any-llm](https://github.com/mozilla-ai/any-llm) library.

Your own personal AI assistant. Any OS.

OpenClaw v2026.3.24 adds OpenAI-compatible endpoints, Teams streaming, a --container flag, Discord auto-thread naming, and a revamped Skills UI.

$ git clone --branch v2026.3.24 https://github.com/openclaw/openclaw.git
# already have the repo? check out this version:
$ git checkout v2026.3.24

$ openclaw --container my-openclaw-container skills list

$ curl http://localhost:3000/v1/models -H 'Authorization: Bearer <api-key>'

$ openclaw skills info openai-whisper-api

• ›Adds /v1/models and /v1/embeddings Gateway endpoints and forwards explicit model overrides through /v1/chat/completions and /v1/responses for OpenAI-compatible client and RAG pipeline support.
• ›Adds --container flag and OPENCLAW_CONTAINER env var to run openclaw commands inside a running Docker or Podman OpenClaw container.
• ›Adds before_dispatch plugin hook with canonical inbound metadata, routing handled replies through the normal final-delivery path and preserving TTS and routed delivery semantics.
• ›Enhances /tools endpoint to show tools the current agent can actually use right now, with a compact default view, optional detailed mode, and a live 'Available Right Now' section in the Control UI.
• ›Migrates Microsoft Teams integration to the official Teams SDK with streaming 1:1 replies, welcome cards with prompt starters, feedback/reflection, typing indicators, status updates, and native AI labeling.

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 v2.1.83 adds drop-in policy fragments, new hook events, credential scrubbing, transcript search, and agent initialPrompt support.

$ git clone --branch v2.1.83 https://github.com/anthropics/claude-code.git
# already have the repo? check out this version:
$ git checkout v2.1.83

$ CLAUDE_CODE_SUBPROCESS_ENV_SCRUB=1 claude

# In managed-settings.json or a managed-settings.d/ fragment:
{ "disableDeepLinkRegistration": true }

• ›Adds managed-settings.d/ drop-in directory so separate teams can deploy independent policy fragments that merge alphabetically alongside managed-settings.json.
• ›Adds CwdChanged and FileChanged hook events for reactive environment management (e.g., direnv integration).
• ›Adds sandbox.failIfUnavailable setting to exit with an error when sandbox is enabled but cannot start, instead of silently running unsandboxed.
• ›Adds disableDeepLinkRegistration setting to prevent claude-cli:// protocol handler registration.
• ›Adds CLAUDE_CODE_SUBPROCESS_ENV_SCRUB=1 environment variable to strip Anthropic and cloud provider credentials from subprocess environments (Bash tool, hooks, MCP stdio servers).

Continue v1.2.18 adds Tensorix and MiniMax LLM providers, Bedrock API key auth, CLI skill importing, and session export/import.

$ git clone --branch v1.2.18-vscode https://github.com/continuedev/continue.git
# already have the repo? check out this version:
$ git checkout v1.2.18-vscode

$ cn review

• ›Adds feat(bedrock): support Bedrock API key authentication — enables API key-based auth for the AWS Bedrock provider.
• ›Adds feat(cli): discover .continue/checks/ directory in cn review for custom review checks.
• ›Adds feat(cli): invokable skills and import any skill — CLI can now invoke and import arbitrary skills.
• ›Adds feat(cli): export and import sessions — persist and restore CLI sessions.
• ›Adds Tensorix as a new LLM provider.

Glamourous agentic coding for all

Crush v0.52.0 adds project-level skills, agent-detection env vars, and smarter file picker ranking.

$ git clone --branch v0.52.0 https://github.com/charmbracelet/crush.git
# already have the repo? check out this version:
$ git checkout v0.52.0

$ if [ "$CRUSH" = "1" ] || [ "$AI_AGENT" = "crush" ]; then echo 'Running inside Crush — skipping interactive prompts'; fi

• ›Adds project-scoped skills: Crush now auto-loads skills from .agents/skills, .crush/skills, .claude/skills, or .cursor/skills, making team skill-sharing as simple as committing files.
• ›Sets CRUSH=1, AGENT=crush, and AI_AGENT=crush environment variables when running Bash commands, letting tools detect they are running inside Crush.
• ›Improves @-triggered file picker with a significantly better completion ranking algorithm, especially useful in large repositories.
• ›Expands Windows global skills path to include ~/.config/crush/skills and ~/.config/agents/skills alongside the AppData location.
• ›Improves system prompt to enforce model responses and session titles in the same language as the user's prompt.

Zed v0.229.0 adds BYOK 1M-context Claude support, editor::AlignSelections, configurable git panel badges, and keymap unbinding via keymap.json.

$ git clone --branch v0.229.0 https://github.com/zed-industries/zed.git
# already have the repo? check out this version:
$ git checkout v0.229.0

[
  {
    "context": "Editor && edit_prediction",
    "unbind": {
      "tab": "editor::AcceptEditPrediction"
    }
  }
]

{
  "git_panel": {
    "show_count_badge": true,
    "file_icons": true,
    "folder_icons": true
  }
}

{
  "label": "Lint current file",
  "command": "lint-runner --lang $ZED_LANGUAGE ${ZED_FILE}"
}

• ›Adds git_panel.show_count_badge config key (off by default) to display a numeric badge on the git panel sidebar icon showing the count of uncommitted changes.
• ›Adds git_panel.file_icons config key (off by default) to enable file type icons from icon theme extensions in the git panel, and git_panel.folder_icons to toggle folder icons (on by default).
• ›Adds unbind key support in keymap.json to remove default key bindings without re-declaring other bindings on the same key.
• ›Adds "formatter": "none" setting to explicitly disable the formatter.
• ›Adds project_panel.scrollbar.horizontal_scroll config key to toggle horizontal scrolling in the project panel.

Get up and running with Kimi-K2.6, GLM-5.2, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models.

Ollama v0.18.3 adds direct VS Code integration via GitHub Copilot, letting local models be used inside the editor.

$ git clone --branch v0.18.3 https://github.com/ollama/ollama.git
# already have the repo? check out this version:
$ git checkout v0.18.3

• ›Enables VS Code GitHub Copilot integration so any local or cloud Ollama model can be selected directly inside Visual Studio Code.

llama.cpp b8526 adds support for the codefuse-ai/F2LLM-v2 model.

$ git clone --branch b8526 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b8526

• ›Adds support for the codefuse-ai/F2LLM-v2 model.

LanceDB python-v0.30.2-beta.0 adds parallel remote inserts via multipart write and a progress bar for add().

$ git clone --branch python-v0.30.2-beta.0 https://github.com/lancedb/lancedb.git
# already have the repo? check out this version:
$ git checkout python-v0.30.2-beta.0

• ›Adds a progress bar to the add() method to track data ingestion in real time.
• ›Enables parallel inserts for remote tables via multipart write, improving throughput for large uploads.
• ›Updates the lance dependency to v3.0.1.

$ git clone --branch v0.27.2-beta.0 https://github.com/lancedb/lancedb.git
# already have the repo? check out this version:
$ git checkout v0.27.2-beta.0