Repository navigation
v0.0.1-beta
Pre-release
Pre-release
Changelog
- 417606f add automation server and python client
- 6beb07f add conversation tool call limits
- 7161e3a add deterministic script provider
- e5a64e6 add fixed tool argument bindings
- 6d6c4f2 add opt-in grouped file edits
- 9957ed1 align prompt guidance with tools and working directory
- 8d5d722 align starter configuration with canonical profiles
- 50c77af allow disabling automatic git snapshots
- 73c3568 autosave sessions once input is admitted
- e0d9af9 bound automation recovery and expose task execution diagnostics
- d11695f bound task execution and compaction and report incomplete runs
- 7ec520c build: add linters and tools to Aura Docker image
- cc43fb2 build: remove multi-arch build flags from Dockerfile
- fefa1f1 checkpoint sessions with leases, frozen forks and confined mcp servers
- e2779b4 chore: Improve UX feedback during initialization
- ed1e69d chore: bump Go version and update dependencies
- 2ffb33a chore: bump to go 1.27
- 30889dd chore: exclude dev branch from release workflow
- d99ee2b chore: make Dockerfile user/group IDs configurable
- b95a810 chore: set default uid to 1000 in dockerfile
- ba39125 ci: run jobs on tags and allow beta releases
- 8850afe clarify bash timeouts for long-running checks
- e322346 clarify starter agent descriptions and profile layout
- 0335cea clarify todo creation and summary updates
- abcfc1c classify native llama.cpp provider failures
- cbfc75d classify structured llama.cpp stream errors
- 83f7cc4 commit completed Aura changes in house style
- 239816e correct unavailable tool feedback
- b18b06c create init destination before extracting sample files
- 8986df2 development branch
- 0086f6b dismiss pending simple ui prompts on shutdown
- a236a79 document go verification checkpoints
- 3f9c761 document the automation contract
- 77e1594 enforce injected tool restrictions throughout each turn
- 9c3fedd escape bash rewrite templates in docs
- 9275cf1 feat: allow explicit/implicit skills
- d970620 feat: allow skills to be invoked like commands
- 7188613 feat: improve skill and model handling
- 330678e feat: normalize thinking effort handling
- a75ad81 fix atomic persistence on windows and check both build targets
- 611e5d0 fix compaction recovery without progress
- 4f274a0 fix cross-directory moves within sandbox write grants
- 9e8d146 fix git-managed plugin updates on installed branches
- 49c1ca9 fix ollama native estimation overflow and thinking
- b153ce1 fix plugin selection across task agents and modes
- 8fb348d fix provider API routing for current coding models
- ff14f2c fix pruning of unread tool results
- 6cfd91b fix task agent model override precedence
- 5b5e646 fix task foreach inheritance
- 90b11d1 fix task max_steps and context deadline
- 0f1455e fix tool recovery and record automation transcripts
- a8f4bdb fix tool result budgets and preserve completion receipts
- 0d965ec fix: validate max steps after overrides
- b40900d forward inference settings during compaction
- 3d9967e forward thinking budgets through compatible gateways
- 7155fd7 forward top-k sampling to llama.cpp
- bde8db2 improve file editing and post-hook feedback
- 7193a2f improve interactive rendering, activity and transcript logging
- 714f06d isolate compaction test model discovery
- 2fdffc5 isolate smoke tests from developer configuration
- f9288a0 keep mutation batches ordered
- 3f36f2f never approve pending prompts without a user
- c3c5c7c preserve compaction evidence and reject incomplete summaries
- f4cc98c preserve file edits and accept empty writes
- b1f28e7 preserve foreach templates across scheduled task runs
- f63213e preserve helper settings and structured responses
- d1f986d preserve learned capabilities in model listings
- 55ce130 preserve retained reasoning in llama.cpp requests
- 4b425e6 preserve thinking settings when model capabilities are unknown
- ae98154 preserve tool schema property order
- 4dd475c quote commands passed to rtk rewrite
- 3d7296b refactor tool definitions and plugin logic
- a3e673d refactor use slices.Backward for reverse iteration
- 06e532a refactor: drop yaml format and lint dependencies
- 1267a27 refactor: inline hook messages in Assistant tools
- 1201ab7 require cross compilers and keep task discovery read only
- 43ec9ab resolve OpenAI models through gateway catalogs
- c915992 restrict automatic Go fixes and lint affected packages
- 125e79d retain tool recovery feedback across turns and resume
- b3c72e0 retry malformed llama.cpp tool responses
- 82b3ea9 return local glob paths relative to the working directory
- f2c378a run task cleanup after failure and cancellation
- e1e3ea2 send mcp cancellations and snapshot tool events
- 5f35fc2 separate tool execution status from output admission
- e17c1bc stop automation when input cannot be admitted
- 993c72c support llama-swap and preserve local model capabilities
- 2519523 support uncapped assistant iterations
- fa66b69 tie serve startup to a launcher lease
- 30da474 treat truncated provider streams as failures
- 359df09 trim redundant todo creation guidance
- e06b269 unify invocation selection and task template context
- 157e8ff update submodule