AI Devtools Radar

Issue #2 · Jul 25, 2026 · week of Jul 20–25

Anthropic sets an Aug 17 Workbench sunset, mem0 closes a tenant-escalation hole, LangChain drops xAI's stop param

The radar watched 63 sources across 32 tools this window and published 10 changes; these are the seven worth your attention, ranked. Every item carries the before/after evidence the pipeline captured — expand it before you take our word.

  1. Anthropic API

    The deadline of the week. The legacy Workbench goes away on August 17, and saved prompts, variables, and evals do not carry over. If anything you care about lives there, export it now, not on the 16th.

    deprecationhigh

    Legacy Workbench being sunset on August 17, 2026

    Users will lose access to the legacy Workbench (platform.claude.com/workbench) in the Claude Console. Saved prompts, variables, and evals from the old Workbench are not supported in the updated version. Users need to export any data they want to keep before August 17, 2026.

    Evidence
    +The legacy Workbench (platform.claude.com/workbench) in the Claude Console is being sunset with access ending on August 17, 2026. Saved prompts, variables, and evals are not supported in the updated Workbench. You can export any data you want to keep from the banner and under your Organizational Settings.

    docs.claude.com

  2. Mem0

    A real security fix, not a routine patch: before this, malicious metadata in update() could move a memory into another tenant's scope. If you run mem0 multi-tenant, upgrade first and read the changelog second.

    featurehigh

    Fix critical security issue preventing metadata from overwriting identity fields (user_id, agent_id, run_id, actor_id)

    Prevents unauthorized tenant scope escalation attacks where malicious metadata could move memories to different tenants

    Evidence
    +Stop `update()` metadata from overwriting or injecting `user_id`, `agent_id`, `run_id`, or `actor_id`. These identity fields are immutable after creation, so passing them in `metadata` can no longer move a memory into a different tenant's scope (#6278)

    github.com

  3. Mem0

    mem0's structured LLM config now reads OPENAI_BASE_URL instead of OPENAI_API_BASE. If you point it at a proxy or compatible provider via the old variable, your override silently stops applying after upgrade.

    api

    OpenAIStructuredLLM now reads OPENAI_BASE_URL instead of OPENAI_API_BASE

    Configuration now matches official OpenAI SDK and other OpenAI-compatible providers; users may need to update environment variables

    Evidence
    +Read `OPENAI_BASE_URL` (was `OPENAI_API_BASE`) in `OpenAIStructuredLLM`, matching the official OpenAI SDK's environment variable and the rest of the OpenAI-compatible providers (#6322)

    github.com

  4. LangChain

    langchain-xai drops the stop parameter it never really supported. Code passing stop to xAI models fails on upgrade. A small change, but one you find out about at runtime, not in code review.

    api

    Dropped unsupported `stop` parameter from XAI integration

    Code using the `stop` parameter with XAI models will need to be updated as this parameter is no longer supported.

    Evidence
    +fix(xai): drop unsupported `stop` parameter (#38335)

    github.com

  5. Weaviate

    Weaviate adds a dedicated REST endpoint for near-text search. If you skipped semantic search to avoid a GraphQL or client-SDK dependency, plain REST now covers it.

    apihigh

    New REST near-text search endpoint added: POST /v1/search/{collection}/near-text

    Developers can now use a dedicated REST endpoint for near-text search operations, expanding API capabilities for semantic search use cases.

    Evidence
    +feat(rest): POST /v1/search/{collection}/near-text — REST near-text search endpoint by @g-despot in #12077

    github.com

  6. Anthropic API

    Claude Managed Agents updates no longer require the version field: omit it to update unconditionally, supply it to get 409-on-mismatch concurrency control. Check which behavior your automation actually wants.

    api

    Claude Managed Agents update semantics now make version field optional

    Users can now update agents without providing a version (unconditional update) or supply it for optimistic concurrency control with 409 error on mismatch.

    Evidence
    +The version field is now optional when updating a Claude Managed Agents agent. Supply it for optimistic concurrency (a mismatch returns a 409 error), or omit it to apply the update unconditionally. See Update semantics.

    docs.claude.com

  7. Neon

    Neon ships neon inspect db, a CLI debug command aimed at agent workflows. Worth a try if your agents touch Neon databases and you are still debugging through the dashboard.

    feature

    New CLI debug command 'neon inspect db' shipped for database debugging

    Users can now debug Neon databases directly from CLI and agent integrations, replacing or complementing previous debugging methods

    Evidence
    −Neon is expanding into a backend: Object Storage, Functions, and AI Gateway now in beta
    +Just shipped: Use neon inspect db to debug your Neon database from your agent/CLI

    neon.com · also seen at 1 more source

That's it for this issue. Get the next one by email — meanwhile the directory updates daily and each issue lives here permanently.