v2026.7.9

Text size

v2026.7.9

CLI Local Module Upload, HashiCorp Vault, LLM Gateways, and Aiden UX

The July 2026 weekly v2026.7.9 release adds CLI upload of custom modules from a local directory (--dir), HashiCorp Vault as an external secret source, and early-access OpenAI-compatible LLM gateways in Aiden. It also unlocks more Aiden alert integrations on the platform UX (Datadog, New Relic, Squadcast, and Dynatrace), and improves Persona Agent, Activities, and agent skills authoring.

For StackGen core, variables of type any or any(list) render in HCL Expression mode by default and are flagged as Action Required when they have no default. Policy evaluation shows generation prompts on the Evaluation tab (with a path back to Generation), and several policy and catalog UI issues are fixed.

Explore the sections below to see what's new, enhanced, and fixed in this update.

	Feature	Link
What's New	Custom Modules	CLI Local Module Upload
	Security	HashiCorp Vault External Secret Source
	Aiden	OpenAI-Compatible LLM Gateways
	Integrations	Aiden Alert Source Integrations
What's Enhanced	Aiden	Persona Agent and Activities UX
		Agent Skills Creation UX
	Custom Modules	Any-Type Variable Rendering
	Governance	Policy Generation Prompt on Evaluation
What's Fixed	Governance	Security Policy Evaluation
		Aiden Policy Try Examples Evaluation
	Catalog	Enterprise Catalog Assign Button
	UX	Attributes Side Panel Icons

What's New

CLI Local Module Upload

Click to view

We've added support to upload custom modules from a local directory through the StackGen CLI, using the --dir flag on stackgen upload custom-modules.

Previously, CLI custom module upload required a Git source (repository URL, ref, branch, tag, and related flags). Now, you can upload Terraform module code from a local directory as a blob into StackGen, without first syncing that module into a Git repository.

Key Features

• Local Directory Upload - Pass --dir with the path to your local Terraform module. Do not combine --dir with Git source flags (--repo-url, --ref, --branch, --tag, --subdir, --secret-name).
• Direct Catalog Publish - Publish modules into StackGen when the platform is your module source of truth, instead of requiring a Git-backed import path first.
• Module Source Preference - Generated Terraform can reference modules from the StackGen Registry by default. To vendor modules into the generated Terraform directory instead, set Module Source to Local in preferences.

Why It Matters

Teams that keep modules outside a synced Git workflow can still get them into the StackGen catalog and use StackGen as the registry for those modules.

Read more in Upload Custom Modules and Module Catalog.

HashiCorp Vault External Secret Source

Click to view

Availability

This is the first version of HashiCorp Vault support. Vault configurations vary; test your auth method, secret paths, and scale before production use. Contact support@stackgen.com or your Customer Service Manager if you need guidance.

We've added HashiCorp Vault as an option when connecting an external secret store.

Previously, external secret stores supported cloud secret managers such as AWS Secrets Manager. Now, you can ingest secrets from HashiCorp Vault as well. This is especially useful for on-prem environments that already store credentials in Vault.

Key Features

• Vault Provider - Choose HashiCorp Vault when you add an external secret store connection.
• Shared Across Products - Once configured, Vault-backed secrets can be used from both StackGen core and Aiden.
• Configure in StackGen Core - Set up the Vault connection in StackGen core secret settings. Aiden uses the same connection after it is configured.

Current Constraints

• First-cut support. Additional Vault scenarios and larger-scale testing are still in progress.
• Connection setup is done in StackGen core today, even when Aiden consumes the secrets.

Why It Matters

On-prem teams that already use Vault can keep secrets in Vault and reference them from StackGen and Aiden, instead of copying them into another store.

Read more in Secret Store.

OpenAI-Compatible LLM Gateways

Click to view

Availability

This feature is early access. The configuration UX works, but still needs polish. Contact support@stackgen.com or your Customer Service Manager for enablement and guidance.

We've added support for OpenAI-compatible LLM gateways in Aiden, so you can route model traffic through a gateway that exposes multiple models behind one OpenAI-compatible API.

Previously, Aiden model setup did not include a path for OpenAI-compatible LLM routers or gateways. Now, you can add a gateway (for example OpenRouter), provide the gateway URL and API key, and select from the models that gateway returns.

Key Features

• OpenAI-Compatible Gateways - Connect any LLM gateway that exposes an OpenAI-compatible API.
• Model Discovery - After you save the connection, Aiden lists the models available from that gateway.
• Purpose-Based Model Selection - Choose different models for different purposes (for example tool calling vs general requests), including cheaper or specialized models where that fits the workload.

Current Constraints

• Early access. The setup and management UX still needs improvement.
• Available models and capabilities depend on the gateway you connect. Confirm tool calling and other requirements for your use case.

Why It Matters

Teams that already run an LLM gateway for cost control, model access, or approved provider lists can use that gateway with Aiden instead of wiring each model provider separately.

Aiden Alert Source Integrations

Click to view

We've expanded Aiden alert and observability integrations on the platform Integrations UX, including Aiden 2.0 availability for sources that were already familiar from earlier Aiden experiences.

Previously, some alert sources were available in earlier Aiden surfaces but were not fully unlocked on the current Integrations page. Now, Datadog and New Relic are available in Community Edition, Squadcast is available as an integration, and Dynatrace is available as an integration (Community Edition UI for Dynatrace is still pending).

Key Features

• Datadog - Available from Integrations on Aiden 2.0 for parity with earlier Datadog support.
• New Relic - Unlocked in the Community Edition Integrations UX.
• Squadcast - Available as an alert and incident integration.
• Dynatrace - Available as an integration. Community Edition UI enablement for Dynatrace is still pending.

Current Constraints

• Dynatrace is available as an integration, but it is not available in the Community Edition UI yet.

Why It Matters

SRE and ops teams can connect the alert sources they already use from the current Aiden Integrations page, with clearer Aiden 2.0 coverage for Datadog and related sources.

What's Enhanced

Persona Agent and Activities UX

Click to view

We've enhanced the Persona Agent management UX and the Activities experience in Aiden.

Previously, Persona Agents were shown mainly as tiles, and activity detail was harder to reach from conversation history. Now, agents appear in a table view with more detail and actions, and activities are easier to open from history, from a chat, and from the execution timeline.

Key Features

• Persona Agent Table View - Review agents in a table with more detail (including who created the agent) and clearer actions.
• Per-Agent Activity Detail - See activities for an agent, when to use them, tools in use, and related workflows.
• Conversation History Hover - Hover a chat in conversation history to preview related activities.
• In-Chat Activities - Open a chat to review each activity and its status, then jump into the execution timeline (graph view). Cost and replay behavior for those views is unchanged.

Why It Matters

It is easier to find the right agent, see what it ran, and inspect execution detail without leaving the chat and history flows you already use.

Agent Skills Creation UX

Click to view

We've enhanced the agent skills creation flow so you can create and edit skills from more entry points in the product.

Previously, skills were primarily imported from a source repository (GitHub, GitLab, or Azure DevOps). Now, you can generate a skill from a prompt, import from a local folder, or upload a skill file, then edit the skill in the same screen after generation or import completes.

Key Features

• Generate from Prompt - Enter a prompt and generate skill content from that description.
• Import from Folder - Import skill content from a local directory.
• Upload Skill File - Upload a skill file directly in the creation flow.
• In-Screen Editing - Edit the skill in the same screen after generation or import finishes.
• Existing Source Import - Continue importing skills from GitHub, GitLab, or Azure DevOps as before.

Why It Matters

You can create and refine agent skills from a prompt, a local folder, or a file, not only from a connected SCM repository.

Any-Type Variable Rendering

Click to view

We've enhanced how variables of type any or any(list) render in the attributes experience, and how missing defaults are flagged.

Previously, these open-ended variable types did not default to expression mode, and missing defaults were easier to miss. Now, any and any(list) variables render in HCL Expression mode by default, and variables without a default are flagged as Action Required.

Key Features

• HCL Expression by Default - Variables of type any or any(list) render in HCL Expression mode by default.
• ui-control Override - Override that default only by explicitly defining a ui-control in stackgen.yaml.
• Action Required Flag - Variables of type any or any(list) that do not have a default value are flagged as Action Required.

Why It Matters

Open-ended module inputs open in the right editor by default, and incomplete any values show up as action items before you move on.

Read more in Custom Module Rendering.

Policy Generation Prompt on Evaluation

Click to view

We've enhanced the Policy Evaluation experience so the generation prompt stays visible, and you can return to Generation when you need to edit it.

Previously, it was easy to lose the prompt that produced a policy once you moved into evaluation. Now, generation prompts are visible on the Evaluation tab, and you can navigate from Evaluation back to the Generation tab to edit the prompt.

Key Features

• Prompt on Evaluation Tab - See the policy generation prompt while you evaluate the policy.
• Return to Generation - Move from the Evaluation tab back to the Generation tab to edit the prompt and continue iterating.

Why It Matters

You can review a policy with the authoring prompt in view, then jump back to Generation without rebuilding the workflow by hand.

What's Fixed

Security Policy Evaluation

Click to view

We fixed an issue where policy evaluation did not work correctly for existing security policies.

Previously, evaluating existing security policies could fail or behave incorrectly. Policy evaluation now works correctly for existing security policies.

Aiden Policy Try Examples Evaluation

Click to view

We fixed an issue in Aiden policy testing where Try examples always returned a positive result for both allow and deny conditions.

Previously, running tests from the Try examples dropdown returned a positive outcome for both positive and negative cases. Allow and deny example evaluation now returns the correct result for the condition under test.

Enterprise Catalog Assign Button

Click to view

We fixed an issue where the Assign button used to assign modules to projects in the Enterprise Catalog table did not render correctly.

Previously, the Assign button could fail to display as expected in the Enterprise Catalog table. The Assign button now renders correctly for module-to-project assignment.

Attributes Side Panel Icons

Click to view

We fixed an issue where policy violation and Action Required icons were not fully visible in the attributes side panel.

Previously, those icons could be clipped or hard to see in the attributes panel. Policy violation and Action Required icons are now fully visible in the attributes side panel.