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 v2.1.183 adds IaC destroy guards, deprecated-model warnings, and a new attribution.sessionUrl setting.
└──▷ GET THIS VERSION
$ git clone --branch v2.1.183 https://github.com/anthropics/claude-code.git
# already have the repo? check out this version:$ git checkout v2.1.183
└──▷ TRY IT
List all shorthand keys you can set inline with /configkey=value, so you don't have to guess field names.
$ /config --help
›Blocks destructive git commands (git reset --hard, git checkout -- ., git clean -fd, git stash drop) in auto mode unless you explicitly asked to discard local work, and blocks git commit --amend on commits not made by the agent this session.
›Blocks terraform destroy, pulumi destroy, and cdk destroy in auto mode unless you asked for the specific stack.
›Adds a warning on stderr (in -p print mode) when the requested model is deprecated or auto-updated to a newer model, now also covering models set in agent frontmatter.
›Adds attribution.sessionUrl setting to omit the claude.ai session link from commits and PRs in web and Remote Control sessions.
›Adds /config --help to list all available shorthand keys for /config key=value.
+1 moreshow less
›Changes /config toggle behavior: Enter and Space both change the selected setting; Esc now saves and closes instead of reverting.
└──▷ BREAKING ON UPGRADE
!The /config Esc key no longer reverts changes — it now saves and closes the config panel.
!The startup 'setup issues' line under the logo has been removed; run /doctor or use --debug to see configuration issues.
Pi v0.79.8 adds Mistral prompt caching, OpenRouter Fusion alias, post-compaction token estimates, and selective provider bundling.
└──▷ GET THIS VERSION
$ git clone --branch v0.79.8 https://github.com/earendil-works/pi.git
# already have the repo? check out this version:$ git checkout v0.79.8
›Supports Mistral provider-side prompt caching using the Pi session ID as prompt_cache_key, with cached-token usage and cost accounting.
›Adds openrouter/fusion as a built-in OpenRouter model alias.
›Adds estimated post-compaction token counts to compact results and compaction events so clients can display approximate context reduction.
›Adds @earendil-works/pi-ai/base and @earendil-works/pi-agent-core/base entry points enabling SDK users to register only the provider transports they need, reducing bundle size.
Hermes v0.17.0 adds iMessage via Photon, Raft agent network, background subagents, image editing, WhatsApp Business API, and atomic memory operations.
└──▷ GET THIS VERSION
$ git clone --branch v2026.6.19 https://github.com/NousResearch/hermes-agent.git
# already have the repo? check out this version:$ git checkout v2026.6.19
└──▷ TRY IT
Connect Hermes to iMessage via Photon so it can send and receive messages in blue-bubble threads without any Mac relay.
$ hermes photon login
Run background skill curation (inactivity sweep only) for free, then opt in to LLM-powered consolidation only when you explicitly want it.
$ hermes curator run --consolidate
›Adds iMessage platform plugin via Photon Spectrum — authenticate with hermes photon login and send/receive iMessages with no Mac relay or BlueBubbles bridge required.
›Adds Raft agent network adapter — connect Hermes as an external gateway agent on Raft via a wake-channel bridge, with privacy-by-contract wake payloads carrying only metadata.
›Adds background/async subagents via delegate_task(background=true) — dispatches a subagent that runs independently and re-enters the conversation as a new turn when complete.
›Adds image-to-image editing to image_generate — pass a source image and prompt to edit, transform, or recompose existing images across all supported image providers.
›Adds Automation Blueprints — schedule recurring tasks by answering guided questions instead of writing cron expressions, rendered as forms, slash commands, or agent conversations.
+9 moreshow less
›Adds grok-composer-2.5-fast (Cursor's Composer model) to the xAI OAuth model picker with a full 200k context window, usable via an existing xAI Grok subscription.
›Adds full profile builder to the dashboard — configure model, skills, and MCP servers from the browser without editing config.yaml, with a global multi-profile switcher.
›Adds official WhatsApp Business Cloud API adapter — connect via Meta's first-party hosted API without a QR-scanning bridge process.
›Adds Telegram Bot API 10.1 rich message support — proper formatting, native markup, and cleaner long-message handling, enabled by default with an opt-out.
›Upgrades the memory tool with an operations array for atomic batch add/replace/remove edits evaluated against the final character budget in a single call.
›Adds Skills Hub browser rehaul — connected hubs, Featured section, full skill previews, and a security scan on each skill before installation.
›Adds curator cost optimization — LLM-powered consolidation now requires explicit opt-in (curator.consolidate: true or hermes curator run --consolidate), making routine background curation zero-token by default.
›Hardens dashboard authentication — all token-required endpoints return 401 behind the OAuth gate, websocket auth uses the served dashboard token, and a warning fires when a public_url override is silently rejected.
›Adds desktop app rebindable keyboard shortcuts, native OS notifications with per-type toggles, live subagent watch-windows, VS Code Marketplace theme installation, resizable terminal pane, RTL/bidi auto-detection, and per-thread composer drafts.
OpenAI Agents SDK v0.17.6 adds pre-approval tool input guardrails and SDK-only custom data for tool outputs.
└──▷ GET THIS VERSION
$ git clone --branch v0.17.6 https://github.com/openai/openai-agents-python.git
# already have the repo? check out this version:$ git checkout v0.17.6
›Adds pre-approval tool input guardrails, enabling validation or interception of tool inputs before a tool call is executed.
›Adds SDK-only custom data for tool outputs, allowing developers to attach arbitrary metadata to tool results without affecting the JSON-compatible contract sent to the model.
llama.cpp server now emits X-Accel-Buffering: no on streaming endpoints to prevent Nginx reverse-proxy buffering.
└──▷ GET THIS VERSION
$ git clone --branch b9714 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:$ git checkout b9714
›Adds X-Accel-Buffering: no response header to all streaming endpoints in llama-server, preventing Nginx from buffering streamed output when deployed behind a reverse proxy.
llama.cpp server gains a schema field with input validation for structured output enforcement.
└──▷ GET THIS VERSION
$ git clone --branch b9707 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:$ git checkout b9707
›Adds schema field to the llama.cpp server API for structured output, with validation that includes field-level error messages when constraints are violated.
Phoenix 17.9.0 adds a server-side bash tool for agents, OAuth2 role-resync control, and label filtering for prompts and datasets.
└──▷ GET THIS VERSION
$ git clone --branch arize-phoenix-v17.9.0 https://github.com/Arize-ai/phoenix.git
# already have the repo? check out this version:$ git checkout arize-phoenix-v17.9.0
└──▷ TRY IT
Disable the server-side bash tool in a hardened deployment where arbitrary shell execution is not permitted.
$ export PHOENIX_AGENTS_DISABLE_BASH=true
›Adds PHOENIX_AGENTS_DISABLE_BASH environment variable to disable the server-side bash tool for agents.
›Adds a server-side bash tool for agents, enabling server-executed shell commands within agent workflows.
›Adds ROLE_RESYNC configuration for OAuth2 identity providers to preserve manually assigned role overrides during resync.
›Adds prompt route contexts to the Phoenix Inference (pxi) interface.
›Adds label filtering for prompts and datasets, with label management directly from the list view and a prompt model column.
+2 moreshow less
›Renders every tool call individually with execution summaries in the pxi UI.
›Shows usage counts for dataset and prompt labels in the UI.
Composio core 0.11.0 forwards userId in trigger flows and exposes new schema-resolution and argument-normalization APIs.
└──▷ GET THIS VERSION
$ 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]
└──▷ USE IT
Gracefully handle toolkits with dangling $ref pointers (e.g. GMAIL_FETCH_EMAILS) in custom schema processing without crashing.
Safely normalize model-supplied tool-call arguments that may arrive as a JSON string rather than an object, and surface a typed error when they cannot be resolved.
javascript
import { normalizeToolArguments, ComposioInvalidToolArgumentsError } from '@composio/core';
try {
const args = normalizeToolArguments(modelToolCallInput);
await composio.tools.execute('MY_TOOL', args);
} catch (e) {
if (e instanceof ComposioInvalidToolArgumentsError) {
console.error('Bad tool arguments from model:', e.message);
}
}
›Forwards userId when creating trigger instances so trigger 2FA flows can verify connected account ownership.
›Adds dereferenceJsonSchema optional second argument { onUnresolved?: 'throw' | 'sentinel'; onReplace?: (ref, reason) => void } to control handling of unresolvable $ref pointers, with new type exports UnresolvedRefStrategy, UnresolvedRefReason, and DereferenceJsonSchemaOptions.
›Adds normalizeToolArguments helper (exported from @composio/core) that routes model-supplied tool-call arguments through a single normalization path — object payloads pass through, JSON strings are parsed, empty/null payloads become {}, and unresolvable values throw a typed ComposioInvalidToolArgumentsError.
›Re-exports the telemetry instance from @composio/core as a public export alongside logger, allowing providers to emit aggregate signals without reaching into package internals.
›MastraProvider.wrapTool now opts inputParameters and outputParameters into 'sentinel' mode for dangling $ref resolution, emitting one logger.warn per (toolSlug, ref) pair and a one-shot telemetry event composio.mastra.wrapTool.danglingRef (respects COMPOSIO_DISABLE_TELEMETRY=true).
+1 moreshow less
›Telemetry batch and error sends are now deferred so the SDK returns results and rethrows errors without waiting on telemetry network requests.
└──▷ BREAKING ON UPGRADE
!The legacy composio.tools.createCustomTool(...) in-memory registry API is removed. Use Tool Router custom tools via experimental_createTool, experimental_createToolkit, and composio.create(..., { experimental: { customTools, customToolkits } }) instead.
THE AI TOOLCHAIN