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.
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 v2.1.202 adds dynamic workflow sizing, OTel workflow attributes, and a multi-level /code-review command.
└──▷ GET THIS VERSION
$ git clone --branch v2.1.202 https://github.com/anthropics/claude-code.git
# already have the repo? check out this version:$ git checkout v2.1.202
└──▷ TRY IT
Cap agent count for lighter workflows to reduce cost or context usage.
$ /config
›New 'Dynamic workflow size' setting in /config lets you guide how many agents Claude spawns in dynamic workflows (small/medium/large) — advisory, not a hard cap.
›Adds workflow.run_id and workflow.name OpenTelemetry attributes to telemetry from workflow-spawned agents, enabling full reconstruction of a workflow run from OTel data.
›New /code-review <level> <pr#> command runs a multi-agent PR review at a chosen effort level; /review <pr> reverts to fast single-pass review.
└──▷ BREAKING ON UPGRADE
!/review <pr> now performs a fast single-pass review instead of the multi-agent review; use /code-review <level> <pr#> to get multi-agent review behavior.
OpenHands 1.41 adds tree-sitter semantic chunking, parallel tool calls, time-bounded API keys, and an org admin conversation dashboard.
└──▷ GET THIS VERSION
$ git clone --branch cloud-1.41.0 https://github.com/All-Hands-AI/OpenHands.git
# already have the repo? check out this version:$ git checkout cloud-1.41.0
└──▷ HOW TO FIND IT
Restrict an API key to a specific validity window so it auto-expires after a contractor engagement ends.
📍In the console, go to Settings › API Keys, create or edit a key, and set the 'Not Before' and 'Expires At' fields to define the active window.
Speed up agent tasks that invoke multiple tools by allowing them to run concurrently instead of sequentially.
📍In the console, go to Agent Settings and set 'tool_concurrency_limit' to the desired number of parallel tool calls (e.g., 4).
›Adds semantic file chunking using tree-sitter AST parsing for more accurate code understanding during agent tasks.
›New organization conversation admin dashboard gives org admins visibility into all conversations under their org.
›Workspace state is now snapshotted at session start and archived before deletion, preserving production workspace history.
›Parallel tool calls are now configurable via a tool_concurrency_limit agent setting, enabling concurrent tool execution.
›API keys gain optional active-window controls (not_before and expires_at) for time-bounded access.
+7 moreshow less
›API keys now support an explicit 'unbound' org scope as a first-class option.
›New GET /organizations/{id}/me endpoint exposes the caller's permissions within an organization.
›Enterprise super roles can be assigned via user.role_id with permission fallback for fine-grained access control.
›Repo and branch metadata is now attached to Laminar trace entries for better observability of agent runs.
›Dynamic marketplace support added for plugin registration, enabling third-party plugin discovery.
›SaaS auth now accepts an api_key cookie as a fallback credential alongside header-based auth.
›Jira Data Center integration panel now surfaces member-appropriate guidance instead of the admin setup form for non-admin users.
llama.cpp b9885 enables tiled matmul on AIX, delivering ~2x prompt-processing speed gains for FP32, Q4_0, and Q8_0 models.
└──▷ GET THIS VERSION
$ git clone --branch b9885 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:$ git checkout b9885
›Enables the matmul_tiled path on AIX by reducing stack buffer footprint for A_pack and B_pack, unlocking ~2x gains in prompt-processing speed for FP32, Q4_0, and Q8_0 models on that platform.
Phoenix v17.19.0 adds trace tree search and a consolidated account menu in the side nav.
└──▷ GET THIS VERSION
$ git clone --branch arize-phoenix-v17.19.0 https://github.com/Arize-ai/phoenix.git
# already have the repo? check out this version:$ git checkout arize-phoenix-v17.19.0
›Adds trace tree search in the app UI, enabling practitioners to search within the trace tree view.
›Adds a consolidated account menu in the side navigation bar.
THE AI TOOLCHAIN