You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Version check run §23630943347 detected the following changes since last check (2026-03-26):
Tool
Previous
Latest
Status
MCP Gateway
v0.2.6
v0.2.7
✅ Updated in constants.go
Copilot CLI
1.0.11
1.0.12
ℹ️ Uses latest
Claude Code
2.1.84
2.1.85
ℹ️ Uses latest
Codex
0.116.0
0.117.0
ℹ️ Uses latest
GitHub MCP Server
v0.32.0
v0.32.0
— No change
Playwright MCP
0.0.68
0.0.68
— No change
Playwright Browser
v1.58.2
v1.58.2
— No change
APM
v0.8.5
v0.8.5
— No change
DefaultMCPGatewayVersion has been updated from v0.2.6 → v0.2.7 in pkg/constants/constants.go and make recompile was run (178/178 workflows recompiled).
--alt-screen[=value] — Use the terminal alternate screen buffer (on|off)
--no-alt-screen — Disable the terminal alternate screen buffer
Added flags:
--enable-reasoning-summaries — Request reasoning summaries for OpenAI models
New subcommand:
providers — Custom Model Providers (BYOK)
Subcommand: providers (BYOK — Bring Your Own Key)
Allows using a custom model provider instead of GitHub Copilot's model routing. Key environment variables:
Variable
Purpose
COPILOT_PROVIDER_BASE_URL
API endpoint URL (required to activate BYOK)
COPILOT_PROVIDER_TYPE
openai (default), azure, or anthropic
COPILOT_PROVIDER_API_KEY
API key (optional for local providers like Ollama)
COPILOT_PROVIDER_BEARER_TOKEN
Bearer token (takes precedence over API key)
COPILOT_PROVIDER_WIRE_API
completions (default) or responses
COPILOT_MODEL
Model name (required for BYOK; no default)
Supports: Ollama, vLLM, Foundry Local, Azure OpenAI, Anthropic, and any OpenAI-compatible endpoint.
View providers Help Output
Custom Model Providers (BYOK):
Use your own model provider instead of GitHub Copilot's model routing.
Set the COPILOT_PROVIDER_BASE_URL environment variable to activate BYOK mode.
GitHub authentication is not required when using a custom provider.
# Ollama (local, no API key required)
$ COPILOT_PROVIDER_BASE_URL=(localhost/redacted) \
COPILOT_MODEL=deepseek-coder-v2:16b \
copilot
# Anthropic
$ COPILOT_PROVIDER_TYPE=anthropic \
COPILOT_PROVIDER_BASE_URL=https://api.anthropic.com \
COPILOT_PROVIDER_API_KEY=sk-ant-... \
COPILOT_MODEL=claude-sonnet-4-20250514 \
copilot
Impact Assessment
Risk: Low — additive features, uses latest tag
Notable: BYOK providers feature enables running Copilot CLI against any OpenAI-compatible or Anthropic endpoint — may be useful for gh-aw local/offline testing scenarios
Summary
Version check run §23630943347 detected the following changes since last check (2026-03-26):
latestlatestlatestDefaultMCPGatewayVersionhas been updated fromv0.2.6→v0.2.7inpkg/constants/constants.goandmake recompilewas run (178/178 workflows recompiled).MCP Gateway v0.2.6 → v0.2.7
Released: 2026-03-27 (today)
Key Features
GHEC Tenant Support (feat: GHEC tenant support for proxy and guard URL parsing gh-aw-mcpg#2481, fix: port-safe GHEC detection in deriveAPIFromServerURL gh-aw-mcpg#2484): MCP Gateway now correctly handles GitHub Enterprise Cloud tenants in proxy routing and guard URL parsing, including port-safe GHEC detection. Configure via
GITHUB_SERVER_URLorGITHUB_API_URL.Trusted Users in AllowOnly Guard (feat: add trusted-users list to AllowOnly guard policy for user integrity elevation gh-aw-mcpg#2584): The
allow-onlyguard policy now supports atrusted-userslist for elevated integrity levels — fine-grained control over sensitive operations.Bug Fixes & Improvements
tools/listfor HTTP backends using custom auth headers (e.g., Atlassian MCP).set_variable,upload_release_asset, andsync_fork.apply_tool_labels.View Full Changelog
Full release notes: https://github.com/github/gh-aw-mcpg/releases/tag/v0.2.7
Key merged PRs:
Impact Assessment
ghcr.io/github/gh-aw-mcpg:v0.2.7Copilot CLI 1.0.11 → 1.0.12 (uses
latest)CLI Changes (from help diff)
Removed flags:
--alt-screen[=value]— Use the terminal alternate screen buffer (on|off)--no-alt-screen— Disable the terminal alternate screen bufferAdded flags:
--enable-reasoning-summaries— Request reasoning summaries for OpenAI modelsNew subcommand:
providers— Custom Model Providers (BYOK)Subcommand:
providers(BYOK — Bring Your Own Key)Allows using a custom model provider instead of GitHub Copilot's model routing. Key environment variables:
COPILOT_PROVIDER_BASE_URLCOPILOT_PROVIDER_TYPEopenai(default),azure, oranthropicCOPILOT_PROVIDER_API_KEYCOPILOT_PROVIDER_BEARER_TOKENCOPILOT_PROVIDER_WIRE_APIcompletions(default) orresponsesCOPILOT_MODELSupports: Ollama, vLLM, Foundry Local, Azure OpenAI, Anthropic, and any OpenAI-compatible endpoint.
View providers Help Output
Impact Assessment
latesttagCodex 0.116.0 → 0.117.0 (uses
latest)View Highlights
/root/agent_a) with structured inter-agent messaging (feat: change multi-agent to use path-like system instead of uuids openai/codex#15313, feat: structured multi-agent output openai/codex#15515)/titlecommand works in both classic and app-server TUI modes (feat(tui): add /title terminal title configuration openai/codex#12334)tui_app_serverfeature by default openai/codex#15661)read_file,grep_filesdeprecated handlers removed (feat: drop artifact tool and feature openai/codex#15851)Impact Assessment
latesttagread_fileandgrep_filestool handlers removed (feat: drop artifact tool and feature openai/codex#15851, chore: drop artifacts lib openai/codex#15864)Claude Code 2.1.84 → 2.1.85 (uses
latest)Minor patch release. No public repository; changelog not available.
Impact Assessment
latesttag, patch-level bumpPackage Links
References: