[ca] Update CLI tool versions: MCP Gateway v0.1.11, Claude Code 2.1.72, Copilot 1.0.3, Codex 0.113.0 #20327

v0.1.9 (2026-03-06)

• MCP Protocol upgraded to 2025-11-25: Improved compatibility with latest MCP clients/servers (Resolve isHTTPConnectionError TODO and update MCPProtocolVersion to 2025-11-25 gh-aw-mcpg#1619)
• go-sdk upgraded to v1.4.0: Consolidates protocol constants, removes deprecated HTTPTransport (go-sdk v1.4.0 upgrade + protocol constant consolidation + HTTPTransport removal gh-aw-mcpg#1598)
• Config Schema v0.53.6: Validated against latest spec schema (🔄 chore: update schema URL to v0.53.6 gh-aw-mcpg#1640)
• customSchemas Documentation: New docs for custom schema URLs and domain constraint (docs: add customSchemas documentation and clarify domain uppercase constraint gh-aw-mcpg#1606)
• Code health: centralized duplicate constants, eliminated duplicate RPCMessageInfo construction, extracted Docker helpers (refactor: eliminate duplicate RPCMessageInfo construction and document flag-default pattern gh-aw-mcpg#1567, refactor: eliminate three duplicate code patterns in launcher, config, and server packages gh-aw-mcpg#1585, Refactor: Consolidate micro-files and extract Docker helpers gh-aw-mcpg#1641)
• Docker: ghcr.io/github/gh-aw-mcpg:v0.1.9

v0.1.10 (2026-03-10)

• GitHub DIFC enforcement: Added WASM guards, guard policies, and LabelAgent integration (feat: Add GitHub DIFC enforcement with WASM guards, guard policies, and LabelAgent integration gh-aw-mcpg#1674)
• Guards renamed: User-facing flags renamed to "guards"; WASM baked into gateway image (github-guard: bake WASM into gateway image, align test/release workflows, and rename user-facing flags to "guards" gh-aw-mcpg#1693)
• Routed URL fix: Fixed host advertisement and allow-only key normalization (fix: routed URL host advertisement and allow-only key normalization gh-aw-mcpg#1683)
• Schema v0.55.0: Config validated against schema v0.55.0 (🔄 chore: pin schema URL to gh-aw v0.55.0 gh-aw-mcpg#1658)
• Switch release workflow to actions-rust-lang/setup-rust-toolchain (Switch release workflow Rust setup to actions-rust-lang/setup-rust-toolchain gh-aw-mcpg#1703)
• Docker: ghcr.io/github/gh-aw-mcpg:v0.1.10

v0.1.11 (2026-03-10)

• WASM build fix: Corrected typo (targets: → target:) in release workflow setup-rust-toolchain action input; this was causing WASM compilation to silently fail (fix: use 'target' (singular) for setup-rust-toolchain action input gh-aw-mcpg#1715)
• Schema v0.57.0: Config validated against latest schema v0.57.0 (🔄 chore: update schema URL to v0.57.0 gh-aw-mcpg#1713)
• Docker: ghcr.io/github/gh-aw-mcpg:v0.1.11

New flag detected in 2.1.72

• --brief: Enable SendUserMessage tool for agent-to-user communication (new in 2.1.72, not present in 2.1.69)

No public repository or GitHub release notes available for Claude Code.

No visible changes to CLI help output between 1.0.2 and 1.0.3. Main help, config, and environment topic help are identical. This appears to be an internal/dependency update.

Repository is private — no public GitHub release notes available.

Subcommand Help Analysis (1.0.3)

• config: No changes from 1.0.2
• environment: No changes from 1.0.2
• Model list: Same as 1.0.2 (claude-sonnet-4.6, claude-opus-4.6, gpt-5.4, gemini-3-pro-preview, etc.)

New Features

• request_permissions tool: Built-in tool allowing running turns to request additional permissions at runtime, with new TUI rendering (Add request permissions tool openai/codex#13092, feat(tui) render request_permissions calls openai/codex#14004)
• Plugin marketplace: Curated marketplace discovery, richer plugin/list metadata, install-time auth checks, plugin/uninstall endpoint (feat: Add curated plugin marketplace + Metadata Cleanup. openai/codex#13712, support plugin/list. openai/codex#13540)
• App server streaming: Upgraded app-server command execution with streaming stdin/stdout/stderr plus TTY/PTY support (app-server: Add streaming and tty/pty capabilities to command/exec openai/codex#13640, Add in-process app server and wire up exec to use it openai/codex#14005)
• Web search config: Full tool configuration support (filters, location), not just on/off (Allow full web search tool config openai/codex#13675)
• Permission-profile config: New permission-profile config language with split filesystem/network sandbox policy plumbing (config: add initial support for the new permission profile config language in config.toml openai/codex#13434, sandboxing: plumb split sandbox policies through runtime openai/codex#13439)
• Image generation: Output files saved to current working directory (Enabling CWD Saving for Image-Gen openai/codex#13607)

Bug Fixes

• Fixed auth error handling for cloud requirements fetch — 401s now trigger normal auth-recovery flow (fix: properly handle 401 error in clound requirement fetch. openai/codex#14049)
• Fixed trust bootstrap to avoid running git before project trust is established (fix: avoid invoking git before project trust is established openai/codex#13804)
• Fixed Windows execution edge cases including PTY TerminateProcess handling and sandbox startup cwd validation (Fix inverted Windows PTY TerminateProcess handling openai/codex#13989, app-server: require absolute cwd for windowsSandbox/setupStart openai/codex#13833)
• Fixed plugin startup: curated plugins loaded correctly in TUI sessions (fix(plugin): Also load curated plugins for TUI. openai/codex#14050)
• Hardened network proxy policy parsing: rejects global wildcard (*) domains (fix: reject global wildcard network proxy domains openai/codex#13789)

Chores

• Moved logs to dedicated SQLite DB with timestamps, pruning, and row limits (Add timestamped SQLite /feedback logs without schema changes openai/codex#13645, Add timestamps to feedback log lines openai/codex#13688)
• Published CLI releases to winget for Windows distribution (Codex/winget auto update openai/codex#12943)