Skip to content

Releases: akira-init1/ChannelTerm

ChannelTerm v0.4.0

Choose a tag to compare

@github-actions github-actions released this 18 Oct 16:29

Full Changelog: v0.2.0...v0.4.0

Debug Linux applications with AI while keeping your shared hardware terminal.

Highlights

  • Managed GDB debugging through MCP: breakpoints, stepping, variables, threads, registers, and call stacks.
  • Authenticated serial GDB bridging, including Windows COM ports with GDB running in WSL2.
  • Shared serial console alongside independent Ethernet/TCP debugging with a target-compatible GDB.
  • Bridge inspection with channelterm gdb-list and visible AI-reported debugging activity.
  • Improved shell-history preservation, bridge lease management, and port allocation.
  • Real GDB integration tests, bundled debugging Skills, and updated documentation in nine languages.

Available for Windows, Linux, and macOS on amd64 and arm64.

ChannelTerm v0.2.0

Choose a tag to compare

@github-actions github-actions released this 14 Oct 02:38

Full Changelog: v0.1.1...v0.2.0

ChannelTerm v0.2.0

Share one terminal session between humans and AI. This release adds shared SSH terminals, SFTP/SCP file transfers, and a standalone Agent Skills package.

What’s new

  • Shared SSH terminals: Connect to a remote host and access the same session from additional terminals or MCP clients. The remote PTY resizes with the terminal that opened the connection.
  • SSH authentication and host verification: Sign in with a password or private key, including encrypted keys. Confirm the host fingerprint on first connection; connections are rejected if a previously trusted host key has changed.
  • SFTP and SCP file transfers: Upload and download files or directories, track transfer progress, and optionally verify file contents with SHA-256. SFTP also supports remote file management and staged replacement of existing files.
  • Improved MCP integration: Tool titles, behavior annotations, output schemas, and server instructions help AI clients choose and use tools correctly.
  • Standalone Skills package: The bundled cterm-serial-debug skill guides AI agents through shared serial debugging workflows for MCUs and embedded Linux devices.

Downloads and installation

  • Prebuilt binaries are available for Windows, Linux, and macOS on amd64 and arm64.
  • To install the skill, download channelterm_skills_0.2.0.zip, extract it, and copy the entire cterm-serial-debug directory into your AI client’s Skills directory.
  • The Skills package does not include the ChannelTerm executable or configure the MCP connection.
  • Use the included SHA256SUMS file to verify your downloads.

ChannelTerm v0.1.1

Choose a tag to compare

@github-actions github-actions released this 22 Sep 03:25

Highlights

  • Added channelterm install for per-user installation without administrator privileges.
  • Added channelterm uninstall for safely removing installed commands and installer-managed PATH changes.
  • Added channelterm uninstall --purge for removing configuration, device state, and authentication data after explicit confirmation.
  • Added current-user PATH management on Windows, Linux, and macOS.
  • Added safe configuration initialization: existing valid configuration is preserved, while malformed configuration is reported instead of overwritten.
  • Added install.json to track installed files, build information, checksums, and PATH ownership.
  • Added installed-version comparison and downgrade protection.
  • Improved uninstall cleanup while preserving unknown user files.
  • On Windows, empty ChannelTerm-specific program directories are now removed after uninstall.

Platform behavior

  • Windows: installs to %LOCALAPPDATA%\Programs\ChannelTerm\bin
  • Linux: installs to ~/.local/bin
  • macOS: installs to ~/.local/bin

Linux and macOS remove the installed channelterm executable and cterm symbolic link, but preserve the shared ~/.local/bin directory.

A standard uninstall preserves user configuration and credentials. Use channelterm uninstall --purge only when all ChannelTerm user data should also be removed.

Full Changelog: v0.1.0...v0.1.1

ChannelTerm v0.1.0

Choose a tag to compare

@github-actions github-actions released this 21 Sep 09:54

ChannelTerm v0.1.0

Initial public release of ChannelTerm, a serial-first shared terminal-session tool for humans and AI agents.

Highlights

  • Shared serial terminal sessions
  • Native serial targets such as COM50 and /dev/ttyUSB0
  • Multiple CLI and MCP clients sharing one Session
  • MCP server and terminal tools
  • File and directory transfer over active serial Sessions
  • Operation leases for coordinating terminal commands and file transfers
  • Builds for Windows, Linux, and macOS on amd64 and arm64
  • SHA-256 checksums for release downloads

Getting Started

Download the binary matching your operating system and architecture.

Linux:

chmod +x channelterm-linux-amd64
./channelterm-linux-amd64 version
./channelterm-linux-amd64 list

macOS:

chmod +x channelterm-darwin-arm64
./channelterm-darwin-arm64 version
./channelterm-darwin-arm64 list

Windows:

.\channelterm-windows-amd64.exe version
.\channelterm-windows-amd64.exe list

Verify downloaded files with SHA256SUMS when possible.

Current Scope

Serial is the only implemented Transport in v0.1.0. SSH and Telnet are future directions and are not currently supported.

Known Limitations

  • Release binaries are not code-signed or notarized. Windows SmartScreen or macOS Gatekeeper may display a warning.
  • Serial access may require platform-specific drivers, device permissions, or group membership.
  • File transfer requires a compatible shell and utilities on the connected target.
  • Real serial behavior depends on the connected hardware, driver, and communication settings.

Documentation

See the following documentation for installation, configuration, workflows, and architecture: