A Bubble Tea TUI for browsing and reopening AI coding sessions (Claude Code, OpenCode, Qwen, Codex).

If you want an AI agent to configure this tool for your machine, copy and paste this prompt:

You are configuring Agent Session Manager for this machine. Read the README section "AI Auto-Config Protocol" and execute exactly this workflow: detect platform/shell, run `agent-manager --doctor --json` (or fallback `go run ./cmd/session-manager --doctor --json` when running from source), propose minimal env/config changes only (no source code edits), apply them, then re-run doctor and report final status plus exact diffs.

How to use:

1. Copy the prompt above.
2. Paste it into your AI agent chat in this repository.
3. Let the AI run the protocol, then review its reported diffs and doctor results.

This release is optimized for WSL-first workflows. The session restore path is:

1. Open a new Windows Terminal tab/window
2. Enter WSL shell
3. cd to the session project path
4. Run the restore command for the selected tool

Install once, run anywhere (GitHub source):

Stable release (recommended):

npm i -g github:RunMintOn/SessionLens#v0.1.3
agent-manager

Latest default branch (bleeding edge, may be unstable):

npm i -g github:RunMintOn/SessionLens
agent-manager

No need to cd into this repo after install. On first run, agent-manager auto-downloads the matching binary from GitHub Releases.

If npm is unavailable, use a prebuilt binary from GitHub Releases and place it in your PATH as agent-manager.

go build -o /tmp/session-manager-mvp ./cmd/session-manager
/tmp/session-manager-mvp

• /: enter search input mode
• Esc (in search mode): clear and exit search
• Tab or Left/Right: switch panel focus
• Up/Down or k/j: move cursor
• Enter (left panel): open selected project shell (double press within 2s)
• Enter (right panel): resume selected session
• h: hide selected session
• H: show hidden sessions overlay
• 1..5: source filter (All, Claude, OpenCode, Qwen, Codex)
• q: quit

You can customize launcher behavior without changing source code.

• ASM_TERMINAL_CMD (default: wt)
• ASM_TERMINAL_ARGS (default platform-dependent, space-separated)
• ASM_WSL_ENTRY_CMD (default: wsl.exe)
• ASM_WSL_SHELL (default: zsh -lic)
• ASM_HOST_SHELL_CMD (default platform-dependent, usually sh)
• ASM_HOST_SHELL_ARGS (default platform-dependent, usually -lc)
• ASM_RESTORE_CMD_CLAUDE (default: claude -r {id})
• ASM_RESTORE_CMD_OPENCODE (default: opencode -s {id})
• ASM_RESTORE_CMD_QWEN (default: qwen -r {id})
• ASM_RESTORE_CMD_CODEX (default: codex resume {id})
• ASM_CONFIG_PATH (optional, override config file location)

Template placeholders:

• {id}: session ID (shell-quoted)
• {project}: project path (shell-quoted)

Example:

export ASM_WSL_SHELL="bash -lc"
export ASM_RESTORE_CMD_QWEN="qwen --resume {id}"

Install/bootstrap-specific environment variables:

• ASM_RELEASE_REPO (override GitHub release repo, format OWNER/REPO)
• ASM_RELEASE_TAG (pin a release tag, e.g. v0.1.3)
• ASM_RELEASE_BASE_URL (use custom release mirror base URL)
• ASM_ALLOW_LATEST_FALLBACK=1 (allow fallback to latest release when pinned tag download fails)
• ASM_SKIP_BOOTSTRAP=1 (disable first-run binary download)

Default config file path:

• Linux/WSL/macOS: ~/.config/agent-session-manager/config.json
• Or override with ASM_CONFIG_PATH

Example config:

{
  "terminal_cmd": "wt",
  "terminal_args": [],
  "wsl_entry_cmd": "wsl.exe",
  "wsl_shell": "zsh -lic",
  "host_shell_cmd": "sh",
  "host_shell_args": ["-lc"],
  "restore_cmd_claude": "claude -r {id}",
  "restore_cmd_opencode": "opencode -s {id}",
  "restore_cmd_qwen": "qwen -r {id}",
  "restore_cmd_codex": "codex resume {id}"
}

Priority:

1. ASM_* environment variables
2. Config file
3. Built-in defaults

Run a quick environment and launcher configuration check:

agent-manager --doctor
# or (repo/dev mode): go run ./cmd/session-manager --doctor

Machine-readable report:

agent-manager --doctor --json
# or (repo/dev mode): go run ./cmd/session-manager --doctor --json

The report includes:

• active launcher configuration
• config value sources (env / file / default)
• config file path
• WSL detection
• terminal binary discovery
• Windows interop status
• command availability for configured restore tools
• structured fix suggestions

Print current effective config (after default + file + env merge):

agent-manager --print-effective-config
# or (repo/dev mode): go run ./cmd/session-manager --print-effective-config

Generate starter config:

# dry-run (prints target path + JSON)
agent-manager --init-config

# write file
agent-manager --init-config --write

# overwrite existing file
agent-manager --init-config --write --force

Use this section if an AI agent should configure the tool for a user.

Scope constraints:

• Allowed: user shell/profile env vars and user config file.
• Not allowed: patching source code for per-user setup.
• Not allowed: unrelated system/package changes unless user confirms.

Inputs the AI must use:

• Platform/shell signals (uname -a, $SHELL, $WSL_DISTRO_NAME).
• Doctor outputs (--doctor and --doctor --json).
• Existing user config path and current env vars.

Steps:

1. Detect platform and shell (uname -a, $SHELL, echo $WSL_DISTRO_NAME).
2. Run doctor:

   agent-manager --doctor
   agent-manager --doctor --json

3. Decide minimal changes:
   • Read JSON checks and suggestions.
   • If terminal missing, set ASM_TERMINAL_CMD.
   • If tool command missing, update ASM_RESTORE_CMD_* or user PATH.
   • If shell mismatch, set ASM_WSL_SHELL or ASM_HOST_SHELL_CMD/ARGS.
4. Initialize config template:

   agent-manager --init-config

5. Write config to either:
   • shell profile (~/.zshrc, ~/.bashrc), or
   • ~/.config/agent-session-manager/config.json
6. Re-run doctor and verify all required commands are ok.

Required AI output contract:

• detected platform and shell
• chosen change path (env vars or config file)
• exact changes applied (before/after or patch-style diff)
• final doctor status (ok/fail) and any remaining blockers

Success criteria:

• doctor reports terminal command and restore tool command as available.
• resuming a sample session works from the right panel.

• wt not found: install Windows Terminal or set ASM_TERMINAL_CMD.
• windows interop unavailable: WSL cannot spawn Windows processes in current environment.
• qwen: not found (or similar): ensure command is available in configured shell, or override restore command via ASM_RESTORE_CMD_*.
• bootstrap failed on first run: check network access to GitHub, or pin a valid release tag:
  • ASM_RELEASE_TAG=v0.1.3 agent-manager
  • ASM_ALLOW_LATEST_FALLBACK=1 agent-manager (for bleeding-edge installs only)
  • ASM_RELEASE_BASE_URL=<mirror-url> agent-manager

go test ./...

This repo includes a tag-based release pipeline (.github/workflows/release.yml):

1. Push a tag like v0.1.3.
2. CI builds and uploads:
   • agent-manager-linux-amd64
   • agent-manager-windows-amd64.exe
   • checksums.txt
3. Users can install:
   • stable tag: npm i -g github:RunMintOn/SessionLens#vX.Y.Z
   • latest default branch: npm i -g github:RunMintOn/SessionLens