This release includes 2 breaking changes for platform teams planning a safe upgrade.
✓ No known CVEs patched in this version
Topics
ReleasePort's take
Light signalVersion v1.73.0 introduces a `--json` flag for the alias list command and caps search output at 1 MiB to prevent memory issues.
Why it matters: The new --json flag enables structured CLI output, simplifying automation; capping search_files_content at 1 MiB stops potential OOM crashes in large‑file scenarios.
Summary
AI summaryBroad release touches Technical Changes, What's New, Bug Fixes, and refactor.
Changes in this release
| Type | Severity | Summary | CVE |
|---|---|---|---|
| Security | Medium |
Restricts MCP catalog to OAuth and anonymous-access servers only, removing API key complexity Restricts MCP catalog to OAuth and anonymous-access servers only, removing API key complexity Source: granite4.1:30b@2026-06-03-audit Confidence: low |
— |
| Feature | Medium |
Blocks `enable_remote_mcp_server` until the server is actually connected, eliminating need to re‑ask questions Blocks `enable_remote_mcp_server` until the server is actually connected, eliminating need to re‑ask questions Source: llm_adapter@2026-06-03 Confidence: high |
— |
| Feature | Low |
Adds `--json` flag to `alias list` command for structured output Adds `--json` flag to `alias list` command for structured output Source: llm_adapter@2026-06-03 Confidence: high |
— |
| Feature | Low |
Adds ContextLimit helper to modelinfo for centralized context window handling Adds ContextLimit helper to modelinfo for centralized context window handling Source: llm_adapter@2026-06-03 Confidence: high |
— |
| Performance | Low |
Uses `strings.Builder` for message merging to reduce memory allocations Uses `strings.Builder` for message merging to reduce memory allocations Source: granite4.1:30b@2026-06-03-audit Confidence: low |
— |
| Performance | Low |
Improves `search_files_content` memory handling for symlinks and device files Improves `search_files_content` memory handling for symlinks and device files Source: granite4.1:30b@2026-06-03-audit Confidence: low |
— |
| Bugfix | High |
Prevents memory explosion in `search_files_content` by capping output at 1 MiB and skipping large files Prevents memory explosion in `search_files_content` by capping output at 1 MiB and skipping large files Source: llm_adapter@2026-06-03 Confidence: high |
— |
| Bugfix | Medium |
Fixes Gemini parallel tool responses by coalescing them into a single Content Fixes Gemini parallel tool responses by coalescing them into a single Content Source: llm_adapter@2026-06-03 Confidence: high |
— |
| Bugfix | Medium |
Fixes custom OpenAI provider routing for Responses-only models (gpt-4.1, o-series, gpt-5, Codex) Fixes custom OpenAI provider routing for Responses-only models (gpt-4.1, o-series, gpt-5, Codex) Source: llm_adapter@2026-06-03 Confidence: high |
— |
| Bugfix | Medium |
Fixes MCP catalog retry logic for existing unstarted entries Fixes MCP catalog retry logic for existing unstarted entries Source: llm_adapter@2026-06-03 Confidence: high |
— |
| Bugfix | Medium |
Fixes rollback behavior when MCP server Start is cancelled during OAuth or Tools operations Fixes rollback behavior when MCP server Start is cancelled during OAuth or Tools operations Source: llm_adapter@2026-06-03 Confidence: high |
— |
| Bugfix | Medium |
Fixes conversation caching to exclude failed chat continuations Fixes conversation caching to exclude failed chat continuations Source: llm_adapter@2026-06-03 Confidence: high |
— |
| Bugfix | Low |
Uses `cmd.Context()` instead of `context.Background()` for proper cancellation support Uses `cmd.Context()` instead of `context.Background()` for proper cancellation support Source: granite4.1:30b@2026-06-03-audit Confidence: low |
— |
| Refactor | Low |
Removes command queueing – commands are now sent immediately Removes command queueing – commands are now sent immediately Source: llm_adapter@2026-06-03 Confidence: high |
— |
| Refactor | Low |
Refactors registry operations to reuse single session across digest and pull operations Refactors registry operations to reuse single session across digest and pull operations Source: granite4.1:30b@2026-06-03-audit Confidence: low |
— |
| Refactor | Low |
Updates OpenAI handler to support newer Responses stream event shapes Updates OpenAI handler to support newer Responses stream event shapes Source: granite4.1:30b@2026-06-03-audit Confidence: low |
— |
| Refactor | Low |
Removes empty query truncation from MCP server search, showing all matching servers Removes empty query truncation from MCP server search, showing all matching servers Source: granite4.1:30b@2026-06-03-audit Confidence: low |
— |
Full changelog
This release improves MCP catalog server management, fixes streaming issues with AI providers, and adds memory protection for file search operations.
What's New
- Adds
--jsonflag toalias listcommand for structured output - Adds ContextLimit helper to modelinfo for centralized context window handling
- Blocks
enable_remote_mcp_serveruntil the server is actually connected, eliminating the need to re-ask questions
Improvements
- Removes command queueing - commands are now sent immediately
- Removes empty query truncation from MCP server search, showing all matching servers
- Restricts MCP catalog to OAuth and anonymous-access servers only, removing API key complexity
Bug Fixes
- Fixes Gemini parallel tool responses by coalescing them into a single Content
- Fixes custom OpenAI provider routing for Responses-only models (gpt-4.1, o-series, gpt-5, Codex)
- Fixes memory explosion in
search_files_contentby capping output at 1 MiB and skipping large files - Fixes MCP catalog retry logic for existing unstarted entries
- Fixes rollback behavior when MCP server Start is cancelled during OAuth or Tools operations
- Fixes conversation caching to exclude failed chat continuations
Technical Changes
- Refactors registry operations to reuse single session across digest and pull operations
- Updates OpenAI handler to support newer Responses stream event shapes
- Uses
cmd.Context()instead ofcontext.Background()for proper cancellation support - Uses
strings.Builderfor message merging to reduce memory allocations - Improves search_files_content memory handling for symlinks and device files
What's Changed
- docs: update CHANGELOG.md for v1.72.0 by @docker-read-write[bot] in https://github.com/docker/docker-agent/pull/2974
- docs: document alias list --json flag and failure-safe conversation caching by @aheritier in https://github.com/docker/docker-agent/pull/2977
- refactor: reuse registry session for OCI pulls by @dgageot in https://github.com/docker/docker-agent/pull/2975
- openai: handle newer Responses stream event shapes by @rumpl in https://github.com/docker/docker-agent/pull/2976
- fix(gemini): coalesce parallel tool responses into a single Content by @t-mizumoto1203 in https://github.com/docker/docker-agent/pull/2959
- Don't queue commands by @rumpl in https://github.com/docker/docker-agent/pull/2979
- chore: bump direct Go dependencies by @dgageot in https://github.com/docker/docker-agent/pull/2980
- fix: use cmd.Context() instead of context.Background() by @dgageot in https://github.com/docker/docker-agent/pull/2981
- feat(mcp_catalog): block enable_remote_mcp_server until the server is connected by @trungutt in https://github.com/docker/docker-agent/pull/2973
- fix: prevent memory explosion in search_files_content by @dgageot in https://github.com/docker/docker-agent/pull/2983
- refactor: remove empty query truncation from MCP server search by @dgageot in https://github.com/docker/docker-agent/pull/2984
- fix(providers): route Responses-only models on custom OpenAI providers by @Sayt-0 in https://github.com/docker/docker-agent/pull/2985
- feat: add ContextLimit helper to modelinfo by @dgageot in https://github.com/docker/docker-agent/pull/2982
- test(mcp): fix staticcheck SA5011 nil-pointer errors in oauth_test by @dgageot in https://github.com/docker/docker-agent/pull/2989
- refactor: use strings.Builder for message merging in oaistream by @dgageot in https://github.com/docker/docker-agent/pull/2986
- refactor: restrict mcp_catalog to oauth and none auth only by @dgageot in https://github.com/docker/docker-agent/pull/2988
New Contributors
- @t-mizumoto1203 made their first contribution in https://github.com/docker/docker-agent/pull/2959
Full Changelog: https://github.com/docker/docker-agent/compare/v1.72.0...v1.73.0
Breaking Changes
- Removes command queueing – commands are now sent immediately
- Restricts MCP catalog to OAuth and anonymous-access servers only, removing API key complexity
Weekly OSS security release digest.
The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.
No spam, unsubscribe anytime.
Share this release
Related context
Related tools
Earlier breaking changes
- v1.71.0 Freezes configuration schema v9 and starts v10 as latest version
Beta — feedback welcome: [email protected]