Skip to content

CHANGELOG.md

Latest commit

 

History

History
179 lines (121 loc) · 11 KB

CHANGELOG.md

File metadata and controls

179 lines (121 loc) · 11 KB

Release History

0.1.21-preview (2026-04-09)

  • [#7484] Detect an agent.manifest.yaml in the current directory and prompt to use it when running azd ai agent init.
  • [#7464] Prompt for agent communication protocol (responses or invocations) when using azd ai agent init with local code.
  • [#7415] Filter azd ai agent init prompts to only show locations and models supported for agent scenarios.
  • [#7410] Fix azd ai agent init --project-id when agent.yaml does not contain a model resource.
  • [#7545] Update agent endpoint handling to use the latest Foundry agent service endpoints.
  • [#7538] Fix azd ai agent invoke streaming output to print each SSE data object on its own line.
  • [#7576] Add validation to azd ai agent init to ensure the manifest path points to a file, not a directory.
  • [#7553] Update azd ai agent init to stop writing AZURE_AI_PROJECT_ENDPOINT and AZURE_OPENAI_ENDPOINT to agent.yaml; azd ai agent run now translates AZURE_AI_* env vars to FOUNDRY_* equivalents for local agent processes.
  • [#7596] Reduce noisy output during azd ai agent init by redirecting internal log statements to the debug log file; verbose details are now only visible when --debug is used.
  • [#7607] Fix azd ai agent init container resource selection to save the chosen CPU/memory tier into agent.yaml and pre-select the existing tier on reruns; remove stale replica mention from post-init message.
  • [#7589] Fix azd ai agent init to prompt for an existing Foundry project when the agent manifest contains no model resources.

0.1.20-preview (2026-04-02)

  • [#7422] Add /invocations API support to azd ai agent invoke, enabling agents to accept arbitrary input passed directly to the agent.
  • [#7341] Fix azd ai agent init writing unnecessary scale configuration for vnext-hosted agents, which is now skipped when vnext is enabled. Thanks @spboyer for the contribution!

0.1.19-preview (2026-03-31)

  • [#7327] Fix azd ai agent init reruns to reuse an existing azd environment instead of failing when a previous attempt already created it.
  • [#7332] Improve azd ai agent init performance when discovering existing Foundry projects, including faster --project-id validation.
  • [#7355] Update generated Application Insights environment variables to reuse the selected connection and avoid redundant connection creation during deployment.
  • [#7373] Fix postdeploy hook failures in projects without hosted agents so unrelated azd projects no longer error during deploy.

0.1.18-preview (2026-03-23)

  • [#7147] Add azd ai agent init support for initializing from an agent template.

0.1.17-preview (2026-03-20)

  • [#7214] Add ASCII art banner with Foundry branding and version info displayed at extension startup.
  • [#7217] Update container settings to use discrete CPU and memory options, and remove min/max replicas prompts.

0.1.16-preview (2026-03-18)

  • [#7141] Add azd ai agent files command group with upload, download, list, and remove subcommands for managing session-scoped files on hosted agent sandboxes.
  • [#7175] Improve input validation, error handling, and path safety across the extension, including hardened path resolution, sensitive data redaction in error paths, and WebSocket origin validation.

Breaking Changes

  • [#7181] Update azd ai agent show and azd ai agent monitor commands to read agent name and version from the azd service entry, removing the requirement to pass them as parameters.

0.1.15-preview (2026-03-13)

  • [#7080] Fix authentication failures (AADSTS70043/AADSTS700082) for multi-tenant and guest users by using UserTenantId for credential resolution

0.1.14-preview (2026-03-10)

  • [#7026] Add azd ai agent run and azd ai agent invoke commands for running agents locally and invoking agents via a /responses call
  • [#6980] Add --model-deployment parameter to azd ai agent init and fix agent init in copilot/CI scenarios
  • [#6979] Detect and classify auth errors from azd core for improved error telemetry

0.1.13-preview (2026-03-04)

  • [#6957] Fix unmarshal error during azd ai agent init

0.1.12-preview (2026-02-27)

  • [#6892] Fix selected model check during azd ai agent init from code to correctly handle existing versus new model deployments
  • [#6909] Add AZURE_AI_PROJECT_ENDPOINT to default agent environment variables and improve AZURE_AI_MODEL_DEPLOYMENT_NAME env var handling
  • [#6895] Add azd ai agent logs and azd ai agent status commands for viewing agent run logs and deployment status
  • [#6901] Add structured error handling with improved service error mapping for more informative error messages

0.1.11-preview (2026-02-24)

  • [#6828] Add new "init from code" flow allowing users to run azd ai agent init without an existing project, template, or manifest
  • [#6867] Add default model selection for the basic init flow

0.1.10-preview (2026-02-19)

  • [#6749] Add "Choose a different model (all regions)" option during model selection recovery
  • [#6749] Display quota availability info in model deployment prompts
  • [#6749] Improve AZURE_AI_PROJECT_ID and deployment capacity validation

0.1.9-preview (2026-02-05)

  • [#6631] Add support for downloading manifests from public repositories without authentication
  • [#6665] Fix manifest download path handling when path contains slashes
  • [#6670] Simplify azd ai agent init to use --minimal flag, reducing prompts
  • [#6672] Block attempts to use extension with prompt agents (not yet supported)
  • [#6683] Fix panic when parsing agent.yaml files without a template field
  • [#6693] Fix unsafe DefaultAzureCredential usage
  • [#6695] Display agent endpoint as plain text with documentation link instead of clickable hyperlink
  • [#6730] Improve model selection handling when model is unavailable in current region

0.1.8-preview (2026-01-26)

  • [#6611] Statically link the Linux amd64 binary for compatibility with older Linux versions

0.1.6-preview (2026-01-22)

  • [#6541] Add metadata capability
  • [#6541] Support AZD_EXT_DEBUG=true for debugging

0.1.5-preview (2026-01-12)

  • [#6468] Add support for retrieving existing Application Insights connections when using --project-id
  • [#6482] Improve azd ai agent init -m validation

0.1.4-preview (2025-12-15)

  • [#6326] Fix correlation ID propagation and improve tracing for API calls
  • [#6343] Improve azd ai agent init completion message to recommend azd up first
  • [#6344] Rename AI_FOUNDRY_PROJECT_APP_ID environment variable to AZURE_AI_PROJECT_PRINCIPAL_ID
  • [#6366] Fix manifest URL path when branch name contains "/"

0.1.3-preview (2025-12-03)

  • Improve agent service debug logging via AZD_EXT_DEBUG env var and --debug flag

0.1.2-preview (2025-11-20)

  • Update extension name and descriptions
  • Update user facing text to use Microsoft Foundry

0.1.1-preview (2025-11-17)

  • Fix min and max replicas not being set during agent deployment
  • Fix azd show not displaying agent endpoint
  • Polish user prompts and messages

0.1.0-preview (2025-11-14)

  • Apply defaults instead of prompting in event handlers
  • Process model resources as parameters
  • Update env var generation to support multi-agent projects
  • Polish error messages
  • Improve local manifest handling
  • Fix agent playground URL generation
  • Fix panic when container settings is nil

0.0.7 (2025-11-13)

  • Add prompting for container resources
  • Add "preview" label to extension name and command descriptions
  • Show agent playground URL post-deploy
  • Support fetching ACR connections from existing AI Foundry projects
  • Fix environment variable references
  • Improve agent name validation

0.0.6 (2025-11-11)

  • Add support for using existing AI model deployments
  • Add --project-id flag for initializing using existing AI Foundry projects
  • Fix agent definition handling for saved templates

0.0.5 (2025-11-06)

  • Add support for tools
  • Improve defaulting logic and --no-prompt support
  • Fix remote build support

0.0.4 (2025-11-05)

  • Add support for --no-prompt and --environment flags in azd ai agent init
  • Include operation ID in timeout error
  • Fix env vars not being included in agent create request

0.0.3 (2025-11-04)

  • Add support for latest MAML format
  • Fix agent endpoint handling for prompt agents

0.0.2 (2025-10-31)

  • Add --host flag to azd ai agent init
  • Rename host type to azure.ai.agent
  • Store model information in service config
  • Display agent endpoint on successful deploy
  • Improve error handling
  • Fix panic when no default model capacity is returned

0.0.1 (2025-10-28)

  • Initial release