Releases: microsoft/BC-Bench
v0.6.0
Introducing Code Review category into BC-Bench.
Bump max parallel to
Leverage bc artifacts caching to speed up categories that requires container setup
Versions
- GitHub Copilot CLI
1.0.57 - Claude Code
2.1.160 - Microsoft.Dynamics.BusinessCentral.Development.Tools
18.0.37.11445-beta
v0.5.6
Enhance extensible result display
Fix MCP server is blocked due to GITHUB_COPILOT_PROMPT_MODE_WORKSPACE_MCP env var after the latest Copilot CLI uptake.
Versions
- GitHub Copilot CLI
1.0.57 - Claude Code
2.1.160 - Microsoft.Dynamics.BusinessCentral.Development.Tools
18.0.37.11445-beta
v0.5.5
Pinned .NET 8 for tasks that are on v24.
Enforced stronger linter rules for code style.
Improved extensibility and made it easier to onboard new categories into BC-Bench.
Uptake latest bc-eval package, storing data in Kusto and uptake LMchecklist (LLM as judge).
Uptake AL-LSP, and enable docker connection with AL MCP via environment variable.
Switched to org-level token instead of PAT.
Versions
- GitHub Copilot CLI
1.0.57 - Claude Code
2.1.160 - Microsoft.Dynamics.BusinessCentral.Development.Tools
18.0.37.11445-beta
v0.5.3
Improved documentation on how to run experiments.
Removed mini-bc-agent, focus on GitHub Copilot and Claude Code.
Stopped pinning BcContainerHelper, always use the latest version, and fixed the v24 tasks in the pipeline.
Bump GitHub Copilot version to 1.0.39 to include gpt-5.5
Introduce an automated candidate screening pipeline for potential dataset refresh.
Enforce additional ruff rules.
Versions
- GitHub Copilot CLI
1.0.39 - Claude Code
2.1.116 - Microsoft.Dynamics.BusinessCentral.Development.Tools
17.0.33.55542
v0.5.2
Refactored tool usage from parsing log files to using PreToolUse hooks.
Bump Claude Code version referencing https://www.anthropic.com/engineering/april-23-postmortem
Pining BCContainerHelper and uv for pipeline instabilities.
Versions
- GitHub Copilot CLI
1.0.31 - Claude Code
2.1.116 - Microsoft.Dynamics.BusinessCentral.Development.Tools
17.0.33.55542 - BcContainerHelper
6.1.12
v0.5.1
v0.5.0
v0.4.0
Fixed the AL MCP compile tool integration in BC-Bench #592
Improved and adjusted the options and settings for Claude Code and GitHub Copilot.
Repository setup now sparse-checks out only app folders, improving clone performance.
Fixed two PNG file conversions for a dataset entry.
Timeout is extended from 90 mins to 120 mins, due to the compile tool from al mcp
Versions
- GitHub Copilot CLI
1.0.2 - Claude Code
2.1.69 - Microsoft.Dynamics.BusinessCentral.Development.Tools
17.0.33.55542