The AI Toolchain — issue -126, April 15, 2026

Agno v2.5.17 adds per-request GitHub repo targeting and a toggle to disable Claude file citations.

$ git clone --branch v2.5.17 https://github.com/agno-agi/agno.git
# already have the repo? check out this version:
$ git checkout v2.5.17

• ›Adds option to disable Claude file citations (PR #7511).
• ›Allows GitHubConfig repo to be specified per request rather than only at configuration time (PR #7496).

AutoGPT Platform adds MemoryEnvelope metadata model with scoped retrieval and memory hardening.

$ git clone --branch autogpt-platform-beta-v0.6.56 https://github.com/Significant-Gravitas/AutoGPT.git
# already have the repo? check out this version:
$ git checkout autogpt-platform-beta-v0.6.56

• ›Introduces MemoryEnvelope metadata model with scoped retrieval and memory hardening for more structured and reliable agent memory management.

$ git clone --branch autogpt-platform-beta-v0.6.55 https://github.com/Significant-Gravitas/AutoGPT.git
# already have the repo? check out this version:
$ git checkout autogpt-platform-beta-v0.6.55

Build resilient agents.

LangGraph checkpoint 4.0.2 documents LANGGRAPH_STRICT_MSGPACK environment variable for checkpoint security hardening.

$ git clone --branch checkpoint==4.0.2 https://github.com/langchain-ai/langgraph.git
# already have the repo? check out this version:
$ git checkout checkpoint==4.0.2

• ›Documents LANGGRAPH_STRICT_MSGPACK environment variable to control strict MessagePack deserialization security for checkpoints.

OpenAI Agents SDK v0.14.0 ships Sandbox Agents — persistent isolated workspaces with shell, filesystem, memory, snapshots, and hosted-provider backends.

$ git clone --branch v0.14.0 https://github.com/openai/openai-agents-python.git
# already have the repo? check out this version:
$ git checkout v0.14.0

• ›Adds SandboxAgent class (extends Agent) with default_manifest, sandbox instructions, capabilities, and run_as for running agents inside persistent, isolated workspaces.
• ›Adds SandboxRunConfig for per-run sandbox wiring: client creation, live session injection, serialized session resume via SandboxSessionState, manifest overrides, snapshots, and materialization_concurrency limits.
• ›Adds Manifest — a workspace-bootstrap contract covering files, directories, local files, local directories, Git repos, environment variables, users, groups, and mounts.
• ›Adds built-in sandbox capabilities for shell access, filesystem editing and image inspection, skills, memory, and compaction.
• ›Adds UnixLocalSandboxClient for fast local development and DockerSandboxClient for container-isolated runs with image parity.

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 v2.1.110 adds flicker-free TUI fullscreen, mobile push notifications, and distributed trace linking for SDK sessions.

$ git clone --branch v2.1.110 https://github.com/anthropics/claude-code.git
# already have the repo? check out this version:
$ git checkout v2.1.110

$ /tui fullscreen

$ TRACEPARENT=00-4bf92f3577b34da6a3ce929d0e0e4736-00f067aa0ba902b7-01 TRACESTATE=vendor=value claude --headless --print "summarise logs"

• ›Adds /tui command and tui setting — run /tui fullscreen to switch to flicker-free rendering within the same conversation.
• ›Adds /focus command to toggle focus view independently from verbose transcript (previously both were on Ctrl+O).
• ›Adds push notification tool so Claude can send mobile push notifications when Remote Control and 'Push when Claude decides' config are enabled.
• ›Adds autoScrollEnabled config to disable conversation auto-scroll in fullscreen mode.
• ›Adds option to show Claude's last response as commented context in the Ctrl+G external editor, enabled via /config.

Glamourous agentic coding for all

Crush v0.58.0 adds support for OpenCode Zen and OpenCode Go model providers.

$ git clone --branch v0.58.0 https://github.com/charmbracelet/crush.git
# already have the repo? check out this version:
$ git checkout v0.58.0

• ›Adds support for OpenCode Zen and OpenCode Go as model providers.

Adds /ask command and copilot plugin marketplace update to GitHub Copilot CLI v1.0.27

$ git clone --branch v1.0.27 https://github.com/github/copilot-cli.git
# already have the repo? check out this version:
$ git checkout v1.0.27

$ /ask What does the OWASP Top 10 say about injection attacks?

$ copilot plugin marketplace update

• ›Adds /ask command to ask a quick question without affecting conversation history.
• ›Adds copilot plugin marketplace update command to refresh plugin catalogs.
• ›Status bar now shows @files and #issues hints while typing, and /help hint when the slash command picker is open.

Lightweight coding agent that runs in your terminal

Codex CLI gains plugin marketplace install, Ctrl+R history search, memory controls, expanded MCP support, and secure devcontainer/bubblewrap sandboxing.

$ git clone --branch rust-v0.121.0 https://github.com/openai/codex.git
# already have the repo? check out this version:
$ git checkout rust-v0.121.0

$ codex marketplace add https://github.com/<org>/<marketplace-repo>

$ codex marketplace add /path/to/local/marketplace

$ # Inside the Codex TUI composer, press Ctrl+R and type keywords to reverse-search prompt history.

• ›Adds codex marketplace add command to install plugin marketplaces from GitHub, git URLs, local directories, and direct marketplace.json URLs.
• ›Adds Ctrl+R reverse search in the TUI composer for prompt history, plus local recall for accepted slash commands.
• ›Adds TUI and app-server controls for memory mode, memory reset/deletion, and memory-extension cleanup, including a dedicated endpoint to delete memories.
• ›Expands MCP/plugin support with MCP Apps tool calls, namespaced MCP tool registration, supports_parallel_tool_calls opt-in flag, and sandbox-state metadata sent through MCP tool metadata.
• ›Adds realtime and app-server APIs for output modality, transcript completion events, and raw turn item injection.

The open source coding agent.

OpenCode v1.4.5 adds OTLP telemetry export, an experimental question API with OpenAPI spec, and a reusable question handler factory.

$ git clone --branch v1.4.5 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v1.4.5

• ›Exports AI SDK telemetry spans to OTLP trace backends for observability integration.
• ›Exposes the experimental question API schema and OpenAPI spec from @opencode-ai/server.
• ›Provides a reusable question handler factory for building custom question API hosts.
• ›Desktop shell commands now start from the home directory.

$ git clone --branch v1.4.4 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v1.4.4

$ opencode export --sanitize

AI agent toolkit: unified LLM API, agent loop, TUI, coding agent CLI

Pi v0.67.3 adds renderShell: "self" for tool renderers to own their outer shell and live backoff countdowns on auto-retry.

$ git clone --branch v0.67.3 https://github.com/earendil-works/pi.git
# already have the repo? check out this version:
$ git checkout v0.67.3

renderShell: "self"

• ›New renderShell: "self" option for custom and built-in tool renderers lets tools own their outer shell instead of the default boxed shell, enabling stable large previews such as edit diffs.
• ›Interactive auto-retry status now displays a live countdown during backoff periods instead of a static retry delay message.

Qwen Code v0.14.5 adds context-usage exposure in SDK/non-interactive mode, a startup profiler, fork subagent, and LLM-based message rewrite middleware.

$ git clone --branch v0.14.5 https://github.com/QwenLM/qwen-code.git
# already have the repo? check out this version:
$ git checkout v0.14.5

• ›Exposes /context usage data in non-interactive mode and the SDK API, letting practitioners programmatically inspect context consumption.
• ›Adds a startup performance profiler to the CLI for measuring and diagnosing launch-time overhead.
• ›Implements a fork subagent for context sharing in core, enabling subagents to inherit and share context from a parent agent.
• ›Adds /account command to the VS Code IDE companion for displaying account information.
• ›Adds LLM-based message rewrite middleware in the ACP layer, supporting custom prompts to transform messages before they reach the model.

Zed v0.232.2 adds order-independent file search, Focus Follows Mouse, markdown preview search, and new AI/config settings.

$ git clone --branch v0.232.2 https://github.com/zed-industries/zed.git
# already have the repo? check out this version:
$ git checkout v0.232.2

$ zed --dev-container /path/to/project

{
  "project_panel": {
    "sort_order": "unicode"
  }
}

• ›Adds --dev-container CLI flag to automatically open a project in a dev container when .devcontainer/ configuration is present.
• ›Adds sort_order setting to project_panel for case-sensitive sorting; options are default, upper, lower, and unicode.
• ›Adds reasoning_effort setting for custom OpenAI-compatible models.
• ›Adds options to pane::DeploySearch for keymap-driven search initiation.
• ›Adds diff.plus and diff.minus theme selectors in the diff syntax grammar.

llama.cpp b8806 adds initial CUDA backend support for the Q1_0 quantization format.

$ git clone --branch b8806 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b8806

• ›Adds initial CUDA backend support for the Q1_0 quantization format, including dp4a acceleration and AMD MMA compatibility.

$ git clone --branch b8802 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b8802

Phoenix 14.6.0 adds an eval harness for the agent system and a list-detail layout for session turns.

$ git clone --branch arize-phoenix-v14.6.0 https://github.com/Arize-ai/phoenix.git
# already have the repo? check out this version:
$ git checkout arize-phoenix-v14.6.0

• ›Adds an eval harness to the PXI agent system prompt for structured evaluation workflows.
• ›Introduces a list-detail layout for session turns in the app UI, improving navigation of multi-turn conversations.

Weaviate v1.35.17 adds backup/restore for INACTIVE tenants and Google AI Studio API key support in multi2vec-google.

$ git clone --branch v1.35.17 https://github.com/weaviate/weaviate.git
# already have the repo? check out this version:
$ git checkout v1.35.17

• ›Adds support for Google AI Studio API key headers in the multi2vec-google module.
• ›Adds baseURL validation support.
• ›Adds support for compactv2 downgrades in HNSW.
• ›Provides a descriptive error on downgrade paths when a module is not available (backward-compatibility improvement).