The AI Toolchain — issue -098, May 13, 2026

AutoGPT Platform adds Slack and Discord blocks, Markdown chat export, AutoPilot task queue, and smarter workflow triggers.

$ git clone --branch autogpt-platform-beta-v0.6.60 https://github.com/Significant-Gravitas/AutoGPT.git
# already have the repo? check out this version:
$ git checkout autogpt-platform-beta-v0.6.60

• ›Adds Export Chat as Markdown option to the session menu for saving agent conversations.
• ›Adds SendSlackMessageBlock for direct Slack message integration within agent workflows.
• ›Adds smart Discord thread support with a setup-link button, human mentions, and bot-to-bot communication.
• ›Introduces AutoPilot task queue capped at 5 running and 15 in-flight tasks per user for resource management.
• ›Adds 'Trigger On Anything' mode for enhanced flexibility in workflow trigger configuration.

PydanticAI v1.95.0 adds native Tool Search on Anthropic/OpenAI, an Instrumentation capability class, and Gemini 3 structured-output + tool combinations.

$ git clone --branch v1.95.0 https://github.com/pydantic/pydantic-ai.git
# already have the repo? check out this version:
$ git checkout v1.95.0

agent = Agent(
    'openai:gpt-4o',
    capabilities=[NativeTool(...), Instrumentation(...)]
)

• ›Adds native Tool Search support for Anthropic and OpenAI providers, with custom search strategies available on any provider.
• ›Introduces the Instrumentation capability class; the existing Agent(instrument=...) parameter is now deprecated in favour of capabilities=[Instrumentation(...)].
• ›Renames 'built-in tools' to 'native tools'; native tools are now registered via capabilities=[NativeTool(...)]; old fields are deprecated ahead of v2.
• ›Adds local= opt-in parameter for provider-adaptive capability fallback; auto-fallback is deprecated.
• ›Supports combining structured output and tool use together for Gemini 3 models via the Google provider.

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 v2.1.141 adds terminalSequence hooks, HTTPS plugin cloning, workspace identity scoping, and richer agent/rewind controls.

$ git clone --branch v2.1.141 https://github.com/anthropics/claude-code.git
# already have the repo? check out this version:
$ git checkout v2.1.141

$ # In your hook script, emit a bell via terminalSequence:
echo '{"terminalSequence": "\a"}'

$ CLAUDE_CODE_PLUGIN_PREFER_HTTPS=1 claude plugin install <plugin-name>

$ claude agents --cwd /path/to/project

• ›Adds terminalSequence field to hook JSON output, enabling hooks to emit desktop notifications, window titles, and bells without a controlling terminal.
• ›Adds CLAUDE_CODE_PLUGIN_PREFER_HTTPS env var to clone GitHub plugin sources over HTTPS instead of SSH, for environments without a GitHub SSH key.
• ›Adds ANTHROPIC_WORKSPACE_ID env var for workload identity federation, scoping minted tokens to a specific workspace when a federation rule covers more than one.
• ›Adds claude agents --cwd <path> to scope the session list to a specific directory.
• ›Extends /feedback to optionally include sessions from the last 24 hours or 7 days, covering issues that span multiple sessions.

GitHub Copilot CLI v1.0.47 adds named fork sessions, j/k navigation in /diff, and improved --resume for cloud agent sessions.

$ git clone --branch v1.0.47 https://github.com/github/copilot-cli.git
# already have the repo? check out this version:
$ git checkout v1.0.47

• ›The /fork command now accepts an optional name, and forked sessions display their origin in the sessions dialog.
• ›The --resume flag now supports Copilot cloud agent sessions where the agent hasn't pushed any changes to its branch.
• ›The /diff view now supports j/k keys for up/down navigation.
• ›Copilot Max subscribers now see only the models available to their subscription tier.

an open source, extensible AI agent that goes beyond code suggestions - install, execute, edit, and test with any LLM

Goose v1.34.0 adds hooks, agent CRUD, project-based system prompts, a TUI binary, and Linux Vulkan local inference.

$ git clone --branch v1.34.0 https://github.com/block/goose.git
# already have the repo? check out this version:
$ git checkout v1.34.0

• ›Adds hooks support for customizable agent behavior at runtime.
• ›Adds CRUD operations for managing agents via ACP.
• ›Enables projects as backend sources with system prompt injection into sessions.
• ›Supports auto-updating for plugins installed to ~/.agents/plugins.
• ›Adds Linux Vulkan support for local inference workloads.

The open source coding agent.

OpenCode v1.14.49 adds DigitalOcean OAuth/Inference Router, a v2 model/provider API, pinned sessions, and prompt @mention autocomplete.

$ git clone --branch v1.14.49 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v1.14.49

• ›Adds DigitalOcean OAuth and Inference Router as a supported provider.
• ›Introduces a v2 model and provider listing API.
• ›Automatically creates a global opencode.jsonc config when none exists.
• ›Enables customize-opencode by default with a linked full schema.
• ›Adds autocomplete for configured @mentions in prompts.

Qwen Code v0.15.11 adds structured headless output, codegraph PR risk analysis, runtime session sidecars, and standalone archive installs.

$ git clone --branch v0.15.11 https://github.com/QwenLM/qwen-code.git
# already have the repo? check out this version:
$ git checkout v0.15.11

$ qwen --headless --json-schema '{"type":"object","properties":{"severity":{"type":"string"},"summary":{"type":"string"}},"required":["severity","summary"]}' 'Review the following diff for security issues: ...'

# In your Qwen Code settings file
# tools.toolSearch.enabled: false

• ›Adds --json-schema flag to headless mode for structured JSON output constrained to a user-supplied schema.
• ›Adds tools.toolSearch.enabled setting to control tool-search behavior for prefix-caching models.
• ›Adds DASHSCOPE_PROXY_BASE_URL environment variable to route DashScope prompt-cache requests through an API gateway.
• ›Writes a runtime.json sidecar file for active sessions, exposing live session metadata on disk.
• ›Injects traceId and spanId into debug log files for OpenTelemetry (OTel) correlation.

Zed v1.2.3 adds Git Graph remote support, MCP 2025-11-25, new editor actions, and path pasting in the terminal.

$ git clone --branch v1.2.3 https://github.com/zed-industries/zed.git
# already have the repo? check out this version:
$ git checkout v1.2.3

// In settings.json
{
  "go_to_definition_scroll_strategy": "preserve"
}

• ›Adds git::ToggleFillCommitEditor action to expand the commit editor to fill the git panel's available vertical space.
• ›Adds a new preserve option to go_to_definition_scroll_strategy that keeps the cursor at the same vertical position within the viewport when navigating to a definition.
• ›Adds editor: convert to base64 and editor: convert from base64 commands to the command palette.
• ›Adds support for MCP version 2025-11-25.
• ›Git Graph: Adds remote support, enabling visualization of remote branches in the graph.

LocalAI v4.2.4 adds Liquid Audio speech-to-speech model support and Vulkan VRAM budget parsing.

$ git clone --branch v4.2.4 https://github.com/mudler/LocalAI.git
# already have the repo? check out this version:
$ git checkout v4.2.4

• ›Adds Liquid Audio speech-to-speech (s2s) model and assistant mode on the talk page via the realtime interface.
• ›Extends VRAM budget and usage detection to parse output from vulkaninfo, broadening GPU visibility for Vulkan-backed hardware.

llama.cpp server and WebUI now support continuing generation on reasoning models with chain-of-thought preservation.

$ git clone --branch b9133 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b9133

• ›Server and WebUI support continued generation on reasoning models by orchestrating thinking_start_tag / thinking_end_tag pairs around prefilled messages, so the stream parser routes chunks correctly after a resume.
• ›Server rejects reasoning prefill on channel-based templates (e.g. GPT-OSS) with an explicit error, limiting scope to simple thinking-tag templates.

$ git clone --branch b9129 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b9129

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

ollama launch opencode now accepts image inputs when using vision models.

$ git clone --branch v0.23.4 https://github.com/ollama/ollama.git
# already have the repo? check out this version:
$ git checkout v0.23.4

$ ollama launch opencode

• ›Enables image input support for vision models via ollama launch opencode.

$ git clone --branch v0.30.0 https://github.com/ollama/ollama.git
# already have the repo? check out this version:
$ git checkout v0.30.0

Phoenix 15.8.0 adds session population for agent traces and thinking controls for Anthropic and Google in the playground.

$ git clone --branch arize-phoenix-v15.8.0 https://github.com/Arize-ai/phoenix.git
# already have the repo? check out this version:
$ git checkout arize-phoenix-v15.8.0

• ›Adds thinking controls for Anthropic and Google models in the playground UI.
• ›Populates project_sessions for /chat and /summary agent traces, enabling session-level grouping of agent interactions.
• ›Rebalances the pxi tool layout for improved workspace ergonomics.

$ git clone --branch arize-phoenix-v15.7.0 https://github.com/Arize-ai/phoenix.git
# already have the repo? check out this version:
$ git checkout arize-phoenix-v15.7.0

Langfuse v3.174.0 adds email verification on signup, configurable blob-export field groups, and exposes exportSource/exportFieldGroups in the public REST API.

$ git clone --branch v3.174.0 https://github.com/langfuse/langfuse.git
# already have the repo? check out this version:
$ git checkout v3.174.0

• ›Exposes exportSource and exportFieldGroups in the public REST API for blob exports, giving teams programmatic control over what data is included in exports.
• ›Adds configurable exportFieldGroups for events export, letting operators select which field groups are included when exporting blob/event data.
• ›Adds exportFieldGroups DB column with tRPC passthrough as the backend plumbing for the new field-group export feature.
• ›Adds email verification on signup to strengthen account authentication at registration time.
• ›Adds more default views to the v4 traces view in the web UI.

Composio @composio/[email protected] surfaces accountType filter on connectedAccounts.list() and namespaces Shared Connection mutation APIs under experimental.

$ git clone --branch @composio/[email protected] https://github.com/ComposioHQ/composio.git
# already have the repo? check out this version:
$ git checkout @composio/[email protected]

const shared = await composio.connectedAccounts.list({
  accountType: 'SHARED',
  userIds: ['user_creator'],
});

await composio.connectedAccounts.link('user_id', 'auth_config_id', {
  experimental: {
    accountType: 'SHARED',
    aclConfigForShared: { allowAllUsers: true },
  },
});

await experimental_updateAcl(composio, 'ca_abc', { allowAllUsers: true });

• ›Adds accountType: 'SHARED' filter to connectedAccounts.list() so SHARED connections can be listed directly without dropping to the raw client.
• ›Moves accountType and aclConfigForShared options on connectedAccounts.link() and session.authorize() under a single experimental block, signalling that the shape may change in future releases.
• ›Promotes connectedAccounts.updateAcl() off the class to a top-level named export experimental_updateAcl(composio, id, opts), consistent with the existing experimental_createTool / experimental_createToolkit pattern.

LanceDB python-v0.32.0 adds IVF_HNSW_FLAT index, model-backed FTS tokenizers, Enum/Pydantic support, and namespace operations.

$ git clone --branch python-v0.32.0 https://github.com/lancedb/lancedb.git
# already have the repo? check out this version:
$ git checkout python-v0.32.0

import lancedb
from lancedb import ClientConfig

db = lancedb.connect(
    "db://my-project",
    api_key="<api_key>",
    client_config=ClientConfig(user_id="[email protected]"),
)

• ›Adds IVF_HNSW_FLAT vector index type, combining IVF partitioning with HNSW graph search over flat (uncompressed) vectors.
• ›Adds user_id field to ClientConfig for per-user identification in enterprise/cloud connections.
• ›Supports model-backed native FTS tokenizers, enabling language-model-driven tokenization for full-text search indexes.
• ›Supports Enum types in Pydantic-to-Arrow schema conversion, so Python enum fields map correctly to Arrow schemas.
• ›Supports child namespace operations and JSON serialization for LanceDBConnection, enabling nested namespace hierarchies.

$ git clone --branch v0.29.0 https://github.com/lancedb/lancedb.git
# already have the repo? check out this version:
$ git checkout v0.29.0

await table.prewarmData();