Agentic Index: agentic AI change log https://agenticindex.io/change-log Verified changes to agentic AI platforms: new capabilities, integrations, pricing, security and funding. Every entry is dated and sourced from public vendor material. en-us Free to reuse with a visible link to agenticindex.io. Terms: https://agenticindex.io/use-this-data Sat, 26 Sep 2026 12:00:00 GMT 720 https://media.base44.com/images/public/69e4cec10b4126d233ee7ee4/a44c0d34a_generated_image.png Agentic Index: agentic AI change log https://agenticindex.io/change-log Boomi Agent Control Tower now supports adding provider agents at the account level, manually or through AI Catalog. https://agenticindex.io/change-log/boomi agenticindex.io:change:boomi-sfjuge-2026-09-26 Sat, 26 Sep 2026 12:00:00 GMT Observability / auditability Enterprise operations agent Agent builder Multi-agent platform Boomi Agent Control Tower now supports adding provider agents at the account level, manually or through AI Catalog. When connecting Salesforce, Microsoft Copilot, or Snowflake Cortex, customers can choose scheduled Boomi Sync or API Push. API Push lets customers supply metadata and usage metrics that those providers do not return through Boomi Sync.

Why it matters: Organizations can choose between automatic inventory synchronization and supplying the measurements needed for agent oversight. API Push requires customers to implement and maintain the data feed, so monitoring coverage depends on their integration.

Impact: Medium. Verification: Partially Verified. Type: Observability / auditability.

Evidence: Official changelog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Anthropic Claude Code: Claude Code 2.1.283 added exact model version matching through availableModelsMatch and an overriding deniedModels managed setting. https://agenticindex.io/change-log/anthropic-claude-code agenticindex.io:change:anthropic-claude-code-dpdgsd-2026-09-25 Fri, 25 Sep 2026 12:00:00 GMT Human approval / guardrails Coding agent Claude Code 2.1.283 added exact model version matching through availableModelsMatch and an overriding deniedModels managed setting. Interactive sessions on third party providers or with telemetry disabled now start in auto mode when no permission mode is configured. An explicit permissions.defaultMode setting still takes precedence.

Why it matters: Enterprise administrators can prevent newly released model versions from becoming available without review. Teams using the affected session configurations should evaluate the new default permission behavior against their approval policies.

Impact: High. Verification: Verified. Type: Human approval / guardrails.

Evidence: GitHub release

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> LiteLLM launched the public preview of LiteAgents, an SDK that supports Deep Agents, Pydantic AI, Claude Agent SDK, Codex and OpenCode… https://agenticindex.io/change-log/litellm agenticindex.io:change:litellm-k9gz3u-2026-09-25 Fri, 25 Sep 2026 12:00:00 GMT MCP / tool calling / API Agent infrastructure LiteLLM launched the public preview of LiteAgents, an SDK that supports Deep Agents, Pydantic AI, Claude Agent SDK, Codex and OpenCode through a common query interface. Profiles select the runtime while retaining shared tools, MCP connections and model configuration; native runtime options remain available. Optional Temporal integration adds recovery from recorded operations and checkpoints, while simple local agents require neither Temporal nor PostgreSQL. Changing runtimes starts new runs and does not migrate an active native session.

Why it matters: Teams can compare agent runtimes on their own workloads with less integration work and add recovery for longer jobs. Production evaluation should account for preview maturity and ensure that tools causing external changes tolerate retries.

Impact: High. Verification: Verified. Type: MCP / tool calling / API.

Evidence: Official blog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> careCycle expanded its assistant's plan data to every US county across Medicare Advantage and PDP plan types. https://agenticindex.io/change-log/carecycle agenticindex.io:change:carecycle-1f7by8-2026-09-25 Fri, 25 Sep 2026 12:00:00 GMT Agent capability Voice agent Healthcare agent Customer support agent careCycle expanded its assistant's plan data to every US county across Medicare Advantage and PDP plan types. Its rebuilt lookup pipeline delivers benefit answers in two seconds, down from 6.5 seconds, according to the release note. The same release lets Pulse routing use custom fields and customer assignment use customer tags.

Why it matters: Agencies serving multiple regions can assess the assistant against a broader range of member plans and reduce pauses during benefit conversations. Custom routing criteria also let teams reflect their own service assignments without relying only on standard customer attributes.

Impact: Medium. Verification: Verified. Type: Agent capability.

Evidence: Official changelog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Zendesk made dedicated Helpjuice, WordPress, and ReadMe knowledge connectors generally available, with rollout beginning September 25. https://agenticindex.io/change-log/zendesk agenticindex.io:change:zendesk-gghu2w-2026-09-25 Fri, 25 Sep 2026 12:00:00 GMT Integrations Customer support agent Zendesk made dedicated Helpjuice, WordPress, and ReadMe knowledge connectors generally available, with rollout beginning September 25. AI agents can use synced Helpjuice articles, public WordPress.com posts, and ReadMe guides. Private or self hosted WordPress content and ReadMe API references are excluded.

Why it matters: Support teams can reuse content from these systems in AI agent answers through dedicated connectors. Buyers should check content coverage and account rollout before replacing an existing import workflow.

Impact: Medium. Verification: Verified. Type: Integrations.

Evidence: Official documentation

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Replit Agent can now analyze an uploaded dataset or connected data source and display interactive charts directly in chat, using… https://agenticindex.io/change-log/replit agenticindex.io:change:replit-3iddu3-2026-09-25 Fri, 25 Sep 2026 12:00:00 GMT Agent capability Agent builder Replit Agent can now analyze an uploaded dataset or connected data source and display interactive charts directly in chat, using technology from its acquisition of Atta. Users can request a specific chart or ask the agent to visualize findings, with the agent handling queries and analytical steps.

Why it matters: Business teams can investigate data and inspect visual results within the same conversation, reducing the need to hand each follow up question to a separate analyst or charting tool.

Impact: Medium. Verification: Verified. Type: Agent capability.

Evidence: Official changelog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Windmill v1.818.0 counts enabled service accounts as operator seats in self hosted Enterprise deployments, billing each as half a seat… https://agenticindex.io/change-log/windmill agenticindex.io:change:windmill-170vtl-2026-09-25 Fri, 25 Sep 2026 12:00:00 GMT Pricing / packaging Agent infrastructure Agent builder Windmill v1.818.0 counts enabled service accounts as operator seats in self hosted Enterprise deployments, billing each as half a seat regardless of its assigned role. Enabled service accounts also count toward the offline license seat cap, and creation is refused when the cap is reached. The related Cloud Team billing change is separate from this release.

Why it matters: Organizations using dedicated identities for agents and automation need to include those accounts in license forecasts. Reviewing unused enabled accounts before upgrading can help avoid unexpected charges or blocked account creation.

Impact: High. Verification: Verified. Type: Pricing / packaging.

Evidence: GitHub release

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> CodeRabbit CLI 0.8.1 adds cr code handoff to create a cloud Coding Agent task from a local session summary, with an optional plan and… https://agenticindex.io/change-log/coderabbit agenticindex.io:change:coderabbit-1j7zf7-2026-09-25 Fri, 25 Sep 2026 12:00:00 GMT MCP / tool calling / API Coding agent CodeRabbit CLI 0.8.1 adds cr code handoff to create a cloud Coding Agent task from a local session summary, with an optional plan and discovered transcript. Handoff output now identifies the branch, commit, transferred files, and cloud task; scripts parsing the earlier output need adjustment.

Why it matters: Teams with Coding Agent access can move ongoing local work into cloud execution with its working context. Existing CLI automation needs compatibility review because the human readable response format changed.

Impact: Medium. Verification: Partially Verified. Type: MCP / tool calling / API.

Evidence: Official changelog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Exa launched Agent Ultra, its highest effort mode for exhaustive research and building large lists of entities. https://agenticindex.io/change-log/exa agenticindex.io:change:exa-zqk2ll-2026-09-25 Fri, 25 Sep 2026 12:00:00 GMT Agent capability Agent infrastructure Exa launched Agent Ultra, its highest effort mode for exhaustive research and building large lists of entities. It coordinates agents across many sources and supports research runs lasting up to three hours. Users can set spending and duration limits or stop a run early while retaining the results gathered so far.

Why it matters: Research teams can allocate more computation to difficult verification and enrichment tasks while controlling the budget for each run. This offers a configurable option for work where completeness matters more than response speed.

Impact: Medium. Verification: Verified. Type: Agent capability.

Evidence: Official blog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Kiro Crew 0.7.0 added a durable queue for accepted subagent work, Task Runner steps, and workflow calls, restoring queued work after… https://agenticindex.io/change-log/kiro agenticindex.io:change:kiro-1csajh-2026-09-24 Thu, 24 Sep 2026 12:00:00 GMT Workflow orchestration Coding agent Kiro Crew 0.7.0 added a durable queue for accepted subagent work, Task Runner steps, and workflow calls, restoring queued work after Gateway restarts. The release also added AWS Fargate to its optional remote crew preview, reaching tasks through AWS Systems Manager without an inbound network rule. Fargate tasks have a default six hour lifetime and a fixed limit of ten concurrent tasks.

Why it matters: Teams evaluating long running coding workflows can assess recovery from service restarts and use disposable remote workers with bounded lifetimes and concurrency.

Impact: High. Verification: Verified. Type: Workflow orchestration.

Evidence: Official changelog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Atlassian began rolling out five capacity planning tools in Rovo MCP, grouped under read_capacity_planning and write_capacity_planning. https://agenticindex.io/change-log/atlassian agenticindex.io:change:atlassian-ivhtxu-2026-09-24 Thu, 24 Sep 2026 12:00:00 GMT MCP / tool calling / API Enterprise operations agent Customer support agent Atlassian began rolling out five capacity planning tools in Rovo MCP, grouped under read_capacity_planning and write_capacity_planning. Agents can retrieve allocations, search nonproject work, update allocations, and create or update nonproject work. Organization administrators control access to the toolsets in Atlassian Administration.

Why it matters: Teams can connect planning assistants to staffing allocations with separate read and write permissions. Buyers should confirm tenant rollout and explicitly enable the appropriate toolsets before relying on automated planning workflows.

Impact: Medium. Verification: Partially Verified. Type: MCP / tool calling / API.

Evidence: Official changelog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Box introduced Doc Split APIs to divide a PDF into separate documents and save them to a chosen Box folder. https://agenticindex.io/change-log/box agenticindex.io:change:box-18gg8z-2026-09-24 Thu, 24 Sep 2026 12:00:00 GMT MCP / tool calling / API Enterprise operations agent Agent builder Box introduced Doc Split APIs to divide a PDF into separate documents and save them to a chosen Box folder. Manual mode uses explicit page ranges, while smart mode interprets natural language instructions such as splitting at QR code separators or isolating signature pages. The asynchronous API creates a job through POST /2.0/document_splits and exposes its status and generated files through a GET endpoint. Requests use API version 2026.0, and smart splits consume one AI Unit per four pages.

Why it matters: Teams can automate document packet preparation before extraction or review without manually separating PDFs. Buyers should account for AI Unit consumption and verify the returned page groupings before routing sensitive documents.

Impact: High. Verification: Verified. Type: MCP / tool calling / API.

Evidence: Official changelog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Browserbase Agents can pause a run for human input and resume within the same browser session. https://agenticindex.io/change-log/browserbase agenticindex.io:change:browserbase-1jwxbb-2026-09-24 Thu, 24 Sep 2026 12:00:00 GMT Human approval / guardrails Browser / computer-use agent Agent infrastructure Browserbase Agents can pause a run for human input and resume within the same browser session. Developers set a natural language pauseWhen condition, which gives the agent a pause tool and lets it enter PAUSED when the condition is met. The resume API accepts a follow up task, such as continuing after a user supplies an MFA code.

Why it matters: Teams can include human checkpoints in browser workflows without rebuilding the session after each interruption. They should test that their pause conditions reliably identify the situations requiring human input.

Impact: High. Verification: Verified. Type: Human approval / guardrails.

Evidence: Official changelog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Cline Desktop v0.0.35 adds Linux support through x64 deb and rpm packages alongside its existing macOS and Windows builds. https://agenticindex.io/change-log/cline agenticindex.io:change:cline-1fbal2-2026-09-24 Thu, 24 Sep 2026 12:00:00 GMT Deployment / data residency Coding agent Cline Desktop v0.0.35 adds Linux support through x64 deb and rpm packages alongside its existing macOS and Windows builds. The Linux app uses a native GTK folder picker and downloads updates in the background, with installation controlled by the user. An AppImage is not available in this release.

Why it matters: Teams using Linux development machines can evaluate the desktop agent on their existing operating system. Deployment owners should check package compatibility and update permissions before rolling it out across managed workstations.

Impact: Medium. Verification: Verified. Type: Deployment / data residency.

Evidence: GitHub release

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Corti updated Guided Generation to pass user edited, added, and discarded facts with explicit type metadata. https://agenticindex.io/change-log/corti agenticindex.io:change:corti-qgp3j2-2026-09-24 Thu, 24 Sep 2026 12:00:00 GMT Human approval / guardrails Healthcare agent Agent infrastructure Corti updated Guided Generation to pass user edited, added, and discarded facts with explicit type metadata. User edits now take precedence over conflicting transcript content, and discarded facts are excluded from generated notes. User added facts and system supplied context also receive explicit priority.

Why it matters: Clinical application teams should test correction and deletion workflows against this release, especially where generated notes previously reintroduced rejected transcript details.

Impact: Medium. Verification: Verified. Type: Human approval / guardrails.

Evidence: Official changelog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> cubic added a repository policy that allows automatic pull request approval while findings at selected priorities remain unresolved. https://agenticindex.io/change-log/cubic agenticindex.io:change:cubic-1nius2-2026-09-24 Thu, 24 Sep 2026 12:00:00 GMT Human approval / guardrails Coding agent cubic added a repository policy that allows automatic pull request approval while findings at selected priorities remain unresolved. Choosing a priority also allows lower priorities, and cubic publishes new findings before submitting a separate approval. Other approval rules and GitHub branch protections still apply.

Why it matters: Engineering teams can decide which findings require human resolution before approval. Review the permitted severity threshold and use shadow mode to assess the policy before enabling live approvals.

Impact: High. Verification: Verified. Type: Human approval / guardrails.

Evidence: Official changelog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> DBOS Python 3.1.0 added workflow rewind, allowing a workflow in a terminal state to restart from a specified step. https://agenticindex.io/change-log/dbos agenticindex.io:change:dbos-gfcvc0-2026-09-24 Thu, 24 Sep 2026 12:00:00 GMT Workflow orchestration Agent infrastructure DBOS Python 3.1.0 added workflow rewind, allowing a workflow in a terminal state to restart from a specified step. The release also added workflow ID reuse policies and local configuration of Conductor metadata only mode.

Why it matters: Teams operating durable agent workflows gain a recovery option for completed or failed runs. Rewind can repeat downstream messages and actions, so recovery procedures should account for those effects before replaying a run.

Impact: Medium. Verification: Verified. Type: Workflow orchestration.

Evidence: GitHub release

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Dust added controls for credit spend checkpoints on individual agents in Agent Builder and across the workspace in Usage settings. https://agenticindex.io/change-log/dust agenticindex.io:change:dust-17zlgx-2026-09-24 Thu, 24 Sep 2026 12:00:00 GMT Human approval / guardrails Enterprise operations agent Dust added controls for credit spend checkpoints on individual agents in Agent Builder and across the workspace in Usage settings. Admins and managers can configure the gate, which pauses an agent after a message consumes at least 600 LLM token credits and asks whether to continue.

Why it matters: Teams can require a human decision before expensive tasks consume more credits while allowing trusted agents to run without that interruption. Review which agents receive an exemption when setting spending policies.

Impact: High. Verification: Verified. Type: Human approval / guardrails.

Evidence: Official changelog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Tines released Records API v2 with endpoints for reading records and record types, searching records, and aggregating results. https://agenticindex.io/change-log/tines agenticindex.io:change:tines-xug3l0-2026-09-24 Thu, 24 Sep 2026 12:00:00 GMT MCP / tool calling / API Security / SOC agent Agent builder Enterprise operations agent Tines released Records API v2 with endpoints for reading records and record types, searching records, and aggregating results. Requests can select fields by name or ID and include linked cases, child records, and full artifacts in the same call. The API standardizes response and error handling for workflows that retrieve Records data.

Why it matters: Teams building agent workflows against Tines can reduce chained retrieval calls and custom parsing logic. The new version provides a clearer integration surface for retrieving only the operational context a workflow needs.

Impact: Medium. Verification: Verified. Type: MCP / tool calling / API.

Evidence: Official changelog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Voiceflow made live agent handoff available on every plan. https://agenticindex.io/change-log/voiceflow agenticindex.io:change:voiceflow-130hl0-2026-09-24 Thu, 24 Sep 2026 12:00:00 GMT Pricing / packaging Customer support agent Agent builder Voice agent Voiceflow made live agent handoff available on every plan. Chat agents can use the Handoff step to route conversations to Ujet, Genesys, Kustomer, Dixa, Sunshine Conversations for Zendesk, or a custom handoff provider.

Why it matters: Teams on lower plans can include human escalation in a chat deployment without changing their Voiceflow plan to unlock that step. Buyers can assess how the handoff fits their existing contact center and support routing.

Impact: High. Verification: Partially Verified. Type: Pricing / packaging.

Evidence: Official changelog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> LangSmith launched the public beta of smithtune, a CLI and coding agent skill that turns recorded agent trajectories into supervised fine… https://agenticindex.io/change-log/langsmith agenticindex.io:change:langsmith-1m7p56-2026-09-24 Thu, 24 Sep 2026 12:00:00 GMT Workflow orchestration Agent infrastructure LangSmith launched the public beta of smithtune, a CLI and coding agent skill that turns recorded agent trajectories into supervised fine tuning workflows. It prepares and persists training datasets, submits training jobs to Fireworks or Baseten, compares the resulting model with the base model through LangSmith evaluations, and can deploy the selected model. Users need a LangSmith account with agent traces and an API key for a supported training provider.

Why it matters: Teams can test whether their production examples support a specialized agent model without building separate data preparation, training, and evaluation pipelines. Recorded datasets and evaluation comparisons give reviewers a way to inspect gains and regressions before deployment.

Impact: High. Verification: Verified. Type: Workflow orchestration.

Evidence: Official blog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> OpenHands Enterprise added Git Sync for automations. https://agenticindex.io/change-log/openhands agenticindex.io:change:openhands-ebjvxv-2026-09-24 Thu, 24 Sep 2026 12:00:00 GMT Human approval / guardrails Coding agent OpenHands Enterprise added Git Sync for automations. Administrators can synchronize automations with a Git repository, retain version history and backups outside the host, and review updates through pull requests before deployment.

Why it matters: Engineering teams can apply their existing code review process to changes in autonomous workflows. This provides a practical governance mechanism for organizations expanding shared coding agents.

Impact: Medium. Verification: Verified. Type: Human approval / guardrails.

Evidence: Official blog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Activepieces 0.92.0 adds agent creation from a natural language description and an AI Router that chooses a workflow route from a question. https://agenticindex.io/change-log/activepieces agenticindex.io:change:activepieces-lan9qs-2026-09-24 Thu, 24 Sep 2026 12:00:00 GMT Agent capability Agent builder Activepieces 0.92.0 adds agent creation from a natural language description and an AI Router that chooses a workflow route from a question. The release also adds a Deno code sandbox, an MCP activity feed recording client calls, and OpenID Connect support for restricting connector access by email domain.

Why it matters: Teams can build agents with less manual setup and inspect the actions external MCP clients execute. Domain restrictions give administrators an additional control when exposing connected business systems.

Impact: Medium. Verification: Verified. Type: Agent capability.

Evidence: GitHub release

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Baz Planner adds shared review of implementation plans published from coding agents, with comments on passages or the entire plan that… https://agenticindex.io/change-log/baz agenticindex.io:change:baz-1f49bv-2026-09-24 Thu, 24 Sep 2026 12:00:00 GMT Human approval / guardrails Coding agent Baz Planner adds shared review of implementation plans published from coding agents, with comments on passages or the entire plan that survive revisions. Teams can compare plan versions, assign reviewers by repository access, review on mobile and follow plan links from pull request comments.

Why it matters: Engineering teams can resolve design questions before agents implement code and keep review feedback connected to the resulting pull request. Persistent comments and version comparisons make it easier to confirm that requested changes were addressed.

Impact: Medium. Verification: Partially Verified. Type: Human approval / guardrails.

Evidence: Official changelog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Braintrust announced Nitro, a new asynchronous query engine for Brainstore that separates storage reads from compute work to accelerate… https://agenticindex.io/change-log/braintrust agenticindex.io:change:braintrust-ok05u5-2026-09-24 Thu, 24 Sep 2026 12:00:00 GMT Observability / auditability Agent infrastructure Braintrust announced Nitro, a new asynchronous query engine for Brainstore that separates storage reads from compute work to accelerate trace investigations. Nitro is automatically enabled for SaaS and BYOC customers, with support starting at dataplane 2.15 for deployments customers host themselves. Braintrust reports that full text searches ran more than twice as fast in its production comparison.

Why it matters: Teams investigating agent failures can search large trace histories with less waiting, including when coding agents issue many concurrent queries. Existing hosted customers receive the improvement without configuring a new search interface.

Impact: Medium. Verification: Verified. Type: Observability / auditability.

Evidence: Official blog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Cognee v1.6.1 adds a Google Drive OAuth connector, Gmail ingestion, and shared drive synchronization, with Google connectors bundled in… https://agenticindex.io/change-log/cognee agenticindex.io:change:cognee-85mffc-2026-09-24 Thu, 24 Sep 2026 12:00:00 GMT Integrations Agent infrastructure Cognee v1.6.1 adds a Google Drive OAuth connector, Gmail ingestion, and shared drive synchronization, with Google connectors bundled in the SDK. The release also streams graph visualization responses and preserves document metadata in retrieved chunks while restricting provenance to datasets the caller can read.

Why it matters: Teams can populate agent memory from shared Google workspace content with fewer integration steps. Large memory graphs become easier to inspect, and retrieval retains clearer links to source documents.

Impact: Medium. Verification: Verified. Type: Integrations.

Evidence: GitHub release

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Ada added the compare_changeset_metrics MCP tool to compare a change set with the live AI Agent baseline in one request. https://agenticindex.io/change-log/ada agenticindex.io:change:ada-tf2ehf-2026-09-23 Wed, 23 Sep 2026 12:00:00 GMT MCP / tool calling / API Customer support agent Ada added the compare_changeset_metrics MCP tool to compare a change set with the live AI Agent baseline in one request. Given a change set ID and date range, it returns containment, resolution, CSAT, enabled custom metric pass rates, cohort volumes, and absolute and relative differences.

Why it matters: Customer service teams can use connected assistants to assess a rollout against live performance without separately retrieving and aligning each metric, supporting decisions about whether to promote an agent change.

Impact: Medium. Verification: Verified. Type: MCP / tool calling / API.

Evidence: Official changelog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Agno 3.0.11 introduced a shared Knowledge retrieval pipeline with MMR reranking for result diversity and recency reranking to favor newer… https://agenticindex.io/change-log/agno agenticindex.io:change:agno-uyyzyp-2026-09-23 Wed, 23 Sep 2026 12:00:00 GMT MCP / tool calling / API Agent infrastructure Agno 3.0.11 introduced a shared Knowledge retrieval pipeline with MMR reranking for result diversity and recency reranking to favor newer revisions. It deprecates reranker configuration on individual vector databases in favor of the Knowledge layer. The release adds guarded APIs for relocating indexed documentation sources, with migration running as a dry run by default. Typed page tool results expose errors, completeness and truncation, while run outputs and API schemas gain a cancellation_stage field.

Why it matters: Teams can apply retrieval policies consistently across storage backends and distinguish incomplete evidence from complete results before an agent acts. Existing integrations should migrate reranker configuration and can replace brittle cancellation message parsing with explicit status data.

Impact: Medium. Verification: Verified. Type: MCP / tool calling / API.

Evidence: GitHub release

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Base44 moved Apps API and white label integration authentication to personal access tokens and published an October 15, 2026 retirement… https://agenticindex.io/change-log/base44 agenticindex.io:change:base44-1hwysp-2026-09-23 Wed, 23 Sep 2026 12:00:00 GMT MCP / tool calling / API Agent builder Base44 moved Apps API and white label integration authentication to personal access tokens and published an October 15, 2026 retirement date for account API keys. Integrations must replace the api_key header with Authorization: Bearer and a personal access token. Monitoring and Audit Logs APIs continue using workspace API keys.

Why it matters: Teams running scripts or integrations against the Apps API need to migrate credentials before the cutoff to preserve service continuity.

Impact: High. Verification: Partially Verified. Type: MCP / tool calling / API.

Evidence: Official changelog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Cartesia introduced Multilingual Voices, a library of more than 50 voices that support up to 25 languages while retaining the same vocal… https://agenticindex.io/change-log/cartesia agenticindex.io:change:cartesia-pcsges-2026-09-23 Wed, 23 Sep 2026 12:00:00 GMT Agent capability Agent infrastructure Voice agent Cartesia introduced Multilingual Voices, a library of more than 50 voices that support up to 25 languages while retaining the same vocal identity. Customers can keep a voice ID across its supported languages and select a locale in speech generation requests. The offering is available through the voice library and API, with language and accent support varying by voice.

Why it matters: Voice agent teams can expand language coverage without assigning a different speaker identity to each market. Buyers should validate each voice's supported locales and fallback behavior before deploying it across languages.

Impact: Medium. Verification: Verified. Type: Agent capability.

Evidence: Official blog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Chatbase expanded Shopify actions so agents can create orders from WhatsApp, Instagram, Facebook Messenger, Chat Bubble and Help Page… https://agenticindex.io/change-log/chatbase agenticindex.io:change:chatbase-n7jia0-2026-09-23 Wed, 23 Sep 2026 12:00:00 GMT Integrations Customer support agent Chatbase expanded Shopify actions so agents can create orders from WhatsApp, Instagram, Facebook Messenger, Chat Bubble and Help Page conversations, using payment links, cash on delivery or free orders permitted by configured procedures. Agents can prepare exchanges as new orders, with a human approving or rejecting the exchange in Chatbase Helpdesk before it reaches Shopify. The same release adds product search from customer images and order tagging for damaged or missing items and their free replacements.

Why it matters: Support teams can delegate purchases and replacement handling while retaining an explicit approval gate for the new exchange workflow. Buyers should distinguish that gate from other Shopify actions and constrain free replacements with appropriate procedures.

Impact: High. Verification: Verified. Type: Integrations.

Evidence: Official changelog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Clio added Canadian case law and legislation to Vincent for research, drafting, analysis and citation verification. https://agenticindex.io/change-log/clio agenticindex.io:change:clio-18ovuy-2026-09-23 Wed, 23 Sep 2026 12:00:00 GMT Agent capability Enterprise operations agent Clio added Canadian case law and legislation to Vincent for research, drafting, analysis and citation verification. The integrated collection includes more than 567,000 judicial decisions plus federal and provincial legislation across Canada's common law provinces. Quebec and Nunavut legislation remain in progress and are outside the announced completed coverage.

Why it matters: Firms handling Canadian matters can now evaluate Vincent against a broader body of domestic authority within their research workflows. Jurisdictional gaps remain material when deciding whether its coverage fits a firm's practice.

Impact: Medium. Verification: Verified. Type: Agent capability.

Evidence: Official announcement

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Forest's September release lets Claude, Dust and other MCP clients start Forest Workflows through its MCP server. https://agenticindex.io/change-log/forest-admin agenticindex.io:change:forest-admin-17p4b5-2026-09-23 Wed, 23 Sep 2026 12:00:00 GMT Workflow orchestration Enterprise operations agent Agent infrastructure Forest's September release lets Claude, Dust and other MCP clients start Forest Workflows through its MCP server. Decision steps now offer Automatic (Deterministic) execution, evaluating conditions against preceding Get Data steps without model interpretation during execution. Manual, AI assisted and Full AI modes remain available for decisions that need interpretation.

Why it matters: Operations teams can make governed business processes available to external agents while keeping predictable routing under explicit rules. Buyers can distinguish where a model's judgment is necessary from the parts of a process that should behave consistently on every run.

Impact: Medium. Verification: Verified. Type: Workflow orchestration.

Evidence: Official changelog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Harvey added agentic search for connected SharePoint, OneDrive, and Google Drive content. https://agenticindex.io/change-log/harvey agenticindex.io:change:harvey-n5kzn9-2026-09-23 Wed, 23 Sep 2026 12:00:00 GMT Integrations Enterprise operations agent Harvey added agentic search for connected SharePoint, OneDrive, and Google Drive content. Users can have Harvey find relevant documents directly instead of locating and selecting the files themselves.

Why it matters: Legal teams evaluating access to institutional knowledge can assess whether Harvey retrieves the right material from their existing document systems with less manual preparation.

Impact: Medium. Verification: Partially Verified. Type: Integrations.

Evidence: Official changelog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Kilo Code CLI 7.7.9 added a goal tool that lets the agent start or resume a session goal and continue execution after the current turn. https://agenticindex.io/change-log/kilo-code agenticindex.io:change:kilo-code-xu391n-2026-09-23 Wed, 23 Sep 2026 12:00:00 GMT Workflow orchestration Coding agent Kilo Code CLI 7.7.9 added a goal tool that lets the agent start or resume a session goal and continue execution after the current turn. The tool is allowed by default, with ask and deny permission options. The same release added recurring or single session schedules through cron_create, cron_list, and cron_delete; tasks survive session resumption and expire after seven days.

Why it matters: Engineering teams can assess unattended coding work with explicit controls over whether an agent may initiate continued execution. The default permission policy becomes a rollout decision for teams that require human oversight.

Impact: High. Verification: Verified. Type: Workflow orchestration.

Evidence: GitHub release

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Arcade announced a new Snowflake toolkit that lets agents discover warehouses, databases, schemas, tables, and columns, then run queries… https://agenticindex.io/change-log/arcade agenticindex.io:change:arcade-rvnbry-2026-09-23 Wed, 23 Sep 2026 12:00:00 GMT Integrations Agent infrastructure Arcade announced a new Snowflake toolkit that lets agents discover warehouses, databases, schemas, tables, and columns, then run queries using the requesting user's identity. Existing Snowflake grants determine what the agent can access. The toolkit permits read only queries and blocks writes, schema changes, and role switches.

Why it matters: Enterprise teams can give agents warehouse access while retaining existing user permissions and data policies. The read only boundary limits mutation risk, though query compute costs still need operational controls.

Impact: Medium. Verification: Verified. Type: Integrations.

Evidence: Official blog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Cursor launched Rollouts and Security Review for Teams and Enterprise customers. https://agenticindex.io/change-log/cursor agenticindex.io:change:cursor-1ejoau-2026-09-23 Wed, 23 Sep 2026 12:00:00 GMT Agent capability Coding agent Cursor launched Rollouts and Security Review for Teams and Enterprise customers. Rollouts creates editable monitoring plans for pull requests and checks deployment health against logs, metrics, and traces in each environment. It can propose a revert or send a regression to a cloud agent, but does not merge or roll back autonomously. Security Review checks pull requests for exploitable vulnerabilities and reports severity, attack paths, and proposed fixes.

Why it matters: Teams can evaluate automated deployment monitoring and security review within their coding workflow. Rollouts requires source control, deployment, and telemetry connections, while remediation retains a review boundary.

Impact: High. Verification: Verified. Type: Agent capability.

Evidence: Official changelog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Microsoft's September 23 release notes add Outlook actions that let Microsoft 365 Copilot delete, move, copy, and categorize email and… https://agenticindex.io/change-log/microsoft agenticindex.io:change:microsoft-1k0nc7-2026-09-23 Wed, 23 Sep 2026 12:00:00 GMT Human approval / guardrails Enterprise operations agent Microsoft's September 23 release notes add Outlook actions that let Microsoft 365 Copilot delete, move, copy, and categorize email and manage inbox rules and folders through natural language. Copilot requests confirmation before acting on more than five messages or changing inbox rules. The capability covers classic and new Outlook for Windows, web, Mac, and mobile.

Why it matters: Teams evaluating delegated mailbox management gain a defined approval boundary for bulk actions and rule changes. Administrators can assess whether those confirmation requirements fit their operating policies before expanding use.

Impact: High. Verification: Verified. Type: Human approval / guardrails.

Evidence: Official changelog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Prime Intellect made Prime Sandboxes generally available through its CLI and SDK and as part of its reinforcement learning stack. https://agenticindex.io/change-log/prime-intellect agenticindex.io:change:prime-intellect-1fz57x-2026-09-23 Wed, 23 Sep 2026 12:00:00 GMT Deployment / data residency Agent infrastructure Prime Intellect made Prime Sandboxes generally available through its CLI and SDK and as part of its reinforcement learning stack. Each sandbox runs a Linux microVM with its own kernel, supporting Docker Compose, background services and custom Docker images. Accounts start with capacity for 1,024 concurrent sandboxes.

Why it matters: Teams training and evaluating agents can use VM isolation while retaining their existing Docker environment definitions. Native Docker and kernel support broaden the workloads they can test without operating the underlying VM fleet.

Impact: High. Verification: Verified. Type: Deployment / data residency.

Evidence: Official blog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Rootly added workflow actions that create Slack Canvases and update them as incident status, severity, roles, and tasks change. https://agenticindex.io/change-log/rootly agenticindex.io:change:rootly-173zbs-2026-09-23 Wed, 23 Sep 2026 12:00:00 GMT Workflow orchestration SRE / DevOps agent Rootly added workflow actions that create Slack Canvases and update them as incident status, severity, roles, and tasks change. Markdown templates and Liquid variables populate incident data, while managed sections preserve notes written by responders. Customers must reconnect Slack to grant the new canvas permissions.

Why it matters: Incident stakeholders can follow a current summary without interrupting responders for manual updates. Preserving human notes supports collaborative incident records alongside automated content.

Impact: Medium. Verification: Verified. Type: Workflow orchestration.

Evidence: Official changelog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> W&B Weave: W&B Server v0.85.0 adds links in both directions between Weave calls and the agent spans they produce or invoke. https://agenticindex.io/change-log/wandb-weave agenticindex.io:change:wandb-weave-lffuaj-2026-09-23 Wed, 23 Sep 2026 12:00:00 GMT Observability / auditability Agent infrastructure W&B Server v0.85.0 adds links in both directions between Weave calls and the agent spans they produce or invoke. The same release adds a configurable process wide query parallelism budget, allowing deployments to reduce peak memory use when Weave workloads load large text heavy tables by trading off request concurrency.

Why it matters: Agent operators can follow execution across call and span views when investigating behavior. Deployment owners can tune resource use for large analytical workloads instead of relying on the fixed default concurrency.

Impact: Medium. Verification: Partially Verified. Type: Observability / auditability.

Evidence: Official changelog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Agent Zero v2.13 replaces runtime=input with input_remote for remote execution, without a compatibility alias. https://agenticindex.io/change-log/agent-zero agenticindex.io:change:agent-zero-1ci28w-2026-09-23 Wed, 23 Sep 2026 12:00:00 GMT MCP / tool calling / API Multi-agent platform Browser / computer-use agent Coding agent Agent builder Agent Zero v2.13 replaces runtime=input with input_remote for remote execution, without a compatibility alias. It also adds Telegram and WhatsApp slash commands with project aware command catalogs and scoped integration settings, plus a configurable browser JavaScript evaluation timeout.

Why it matters: Existing remote execution workflows need a migration before upgrading. Operators gain more direct control over agents through messaging channels and a recovery mechanism for browser scripts that stall.

Impact: High. Verification: Verified. Type: MCP / tool calling / API.

Evidence: GitHub release

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> depthfirst makes transitive dependency reachability available to customers as part of every software composition analysis scan. https://agenticindex.io/change-log/depthfirst agenticindex.io:change:depthfirst-1c04xl-2026-09-23 Wed, 23 Sep 2026 12:00:00 GMT Agent capability Security / SOC agent Coding agent depthfirst makes transitive dependency reachability available to customers as part of every software composition analysis scan. Its agent follows execution paths through intermediate packages and exposes the chain of calls leading from application code to a vulnerable function. The analysis can follow invocation mechanisms that static call graphs miss, including command line startup and framework callbacks.

Why it matters: Security teams gain evidence for prioritizing vulnerable dependencies that application code can actually reach. Engineers can inspect the reported path before deciding whether a finding warrants remediation.

Impact: Medium. Verification: Verified. Type: Agent capability.

Evidence: Official blog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Intercom introduced custom reporting metrics that can be filtered by Fin AI Agent ratings or used to count escalations from AI agents to… https://agenticindex.io/change-log/intercom-fin agenticindex.io:change:intercom-fin-1ce4a4-2026-09-23 Wed, 23 Sep 2026 12:00:00 GMT Observability / auditability Customer support agent Intercom introduced custom reporting metrics that can be filtered by Fin AI Agent ratings or used to count escalations from AI agents to humans. Teams can define counts, percentages, and ratios with separate filters on each side of a formula. Metrics apply to the full conversation history and are available across workspace charts on Advanced and Expert plans.

Why it matters: Support teams using Fin on Intercom can track their own quality and escalation measures without rebuilding the same calculation in each report. Historical coverage lets them establish a baseline immediately.

Impact: Medium. Verification: Verified. Type: Observability / auditability.

Evidence: Official changelog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> FlowX.AI documented a Browser Automation node for Agent Builder, available from release 5.13.0 on SaaS. https://agenticindex.io/change-log/flowx-ai agenticindex.io:change:flowx-ai-b7s3kq-2026-09-23 Wed, 23 Sep 2026 12:00:00 GMT Agent capability Agent builder Enterprise operations agent Agent infrastructure Multi-agent platform FlowX.AI documented a Browser Automation node for Agent Builder, available from release 5.13.0 on SaaS. The node drives a real browser through web applications that have no API, filling multi page forms, navigating portals and retrieving results from a task written in plain language. It checks outcomes against the target site's recorded state rather than the agent's own report and keeps the browser session across workflow steps. Self hosted deployments receive it with the next LTS release family.

Why it matters: Banks and insurers running agents on FlowX can now reach legacy portals and third party sites that expose no integration point from inside the same governed workflow, with every run logged. Verification against the site's recorded state is the detail to test in a pilot, since an agent reporting success on a form it never submitted is the usual failure in this category.

Impact: High. Verification: Verified. Type: Agent capability.

Evidence: Official documentation

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Factorial Code introduced Factorial Actions, one configuration block that exposes a process through UI buttons, forms, Factorial One tools… https://agenticindex.io/change-log/factorial agenticindex.io:change:factorial-1vn7jm-2026-09-22 Tue, 22 Sep 2026 12:00:00 GMT Human approval / guardrails Enterprise operations agent Factorial Code introduced Factorial Actions, one configuration block that exposes a process through UI buttons, forms, Factorial One tools or backend jobs and declares required user policies. One tool contracts classify effects as READ, WRITE or DESTRUCTIVE: READ runs without asking, WRITE requests approval, and DESTRUCTIVE requests approval with an irreversible action warning; an unset effect is treated as DESTRUCTIVE. Contracts also describe preconditions, usage limits, partial failure behavior and an optional preview process, while the CLI and API preserve legacy form and UI trigger settings.

Why it matters: Teams extending One can define approval boundaries alongside the code that performs business actions. Owners must classify effects accurately and configure required policies, since an empty policy list leaves an exposed action unrestricted for users who can reach it.

Impact: High. Verification: Partially Verified. Type: Human approval / guardrails.

Evidence: Official changelog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> GitHub Copilot for JetBrains 1.18.0 introduced assisted approvals in public preview, automatically approving lower risk tool calls while… https://agenticindex.io/change-log/github-copilot agenticindex.io:change:github-copilot-1w9swd-2026-09-22 Tue, 22 Sep 2026 12:00:00 GMT Human approval / guardrails Coding agent GitHub Copilot for JetBrains 1.18.0 introduced assisted approvals in public preview, automatically approving lower risk tool calls while requesting human decisions for higher risk actions. The release also added persistent controls for individual MCP tools, shared organization skills and instructions, and the ability to rewind a conversation and its file changes when editing an earlier request.

Why it matters: Engineering teams can evaluate finer control over agent permissions and a clearer way to recover when an agent follows the wrong direction. Shared instructions also support consistent operation across developers.

Impact: High. Verification: Verified. Type: Human approval / guardrails.

Evidence: Official changelog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Google Antigravity SDK 0.1.18 adds local model execution through LiteRT and local servers exposing an OpenAI compatible API. https://agenticindex.io/change-log/google-antigravity agenticindex.io:change:google-antigravity-9wntey-2026-09-22 Tue, 22 Sep 2026 12:00:00 GMT Deployment / data residency Coding agent Google Antigravity SDK 0.1.18 adds local model execution through LiteRT and local servers exposing an OpenAI compatible API. The release adds independent subagent model selection and a coding evaluation preset, enables scheduling in standard tool groups, and removes interactive questions from default tools. It also removes the deprecated modified_arguments_json hook field in favor of modified_args and consolidates compaction settings around token_threshold.

Why it matters: Teams can assess running coding workflows without sending model inference to a hosted service, subject to local hardware and model constraints. Existing integrations need to review tool defaults and update deprecated hook and compaction configurations before upgrading.

Impact: High. Verification: Verified. Type: Deployment / data residency.

Evidence: Official documentation

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Hex now lets Explorer users ask follow up questions inside published Generative Apps and continue investigating in an agent Thread. https://agenticindex.io/change-log/hex agenticindex.io:change:hex-imv6c-2026-09-22 Tue, 22 Sep 2026 12:00:00 GMT Agent capability Data analyst agent Hex now lets Explorer users ask follow up questions inside published Generative Apps and continue investigating in an agent Thread. The same release adds CSV attachments to scheduled Generative App notifications. Exports use underlying project cell outputs, excluding transformations applied only in the app interface.

Why it matters: Business users can pursue new questions from an existing dashboard without rebuilding its context in a separate workflow. Teams distributing scheduled data should review the underlying outputs against recipient permissions, since visible app filters do not determine the exported dataset.

Impact: Medium. Verification: Verified. Type: Agent capability.

Evidence: Official changelog

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]> Mastra released @mastra/mcp 2.0.0, whose servers support only the July 28, 2026 MCP revision and require @mastra/core 1.68 or newer. https://agenticindex.io/change-log/mastra agenticindex.io:change:mastra-4khz10-2026-09-22 Tue, 22 Sep 2026 12:00:00 GMT MCP / tool calling / API Agent infrastructure Mastra released @mastra/mcp 2.0.0, whose servers support only the July 28, 2026 MCP revision and require @mastra/core 1.68 or newer. Tools that need caller input now suspend and resume through input_required continuations with signed request state. The release removes legacy initialization, session headers, standalone HTTP plus SSE transport, and earlier elicitation APIs.

Why it matters: Teams upgrading existing MCP integrations need to revise transport and input handling code and verify client compatibility. Deployments serving older clients can retain supported 1.x servers alongside 2.x servers.

Impact: High. Verification: Verified. Type: MCP / tool calling / API.

Evidence: GitHub release

Source: Agentic Index change log. Free to reuse with a link back. Terms.

]]>