Projects were pinned to the image they were first created from. Both
create paths preferred triple-c-snapshot-<id>:latest whenever it
existed, and container_needs_recreation compared the container's live
image against the triple-c.image label — which create_container wrote
from the same image it created from. A tautology that could never fire.
The only escape was Reset, which calls remove_project_volumes and
destroys the login, skills and transcripts.
Measured consequences on this host: real projects are missing socat (so
the auth bridge cannot tunnel) and bubblewrap (so sandbox mode does not
work), plus Mission Control and triple-c-sso-refresh, and sit 61
packages behind the base including ca-certificates, openssl and curl.
Detection. create_container now writes triple-c.base-image-id (the image
ID, not RepoDigests, which local-built and custom images do not have)
and triple-c.create-image. container_needs_recreation takes the expected
create-image and compares against the latter, so the check means
something. base-image-id is deliberately NOT compared: a base bump would
otherwise silently recreate from the snapshot, consuming the "you should
migrate" signal without migrating. Staleness is surfaced, never acted on
automatically.
Migration keeps the volumes. /home/claude and ~/.claude are volumes and
the image's copy is seed-only — permanently masked after first mount —
so the login, ~/.claude.json, skills, transcripts, scheduler tasks, SSH
keys, cargo, uv, ruff and Claude Code itself re-attach untouched. Only
root-level state is rebuilt: apt packages are replayed against the new
base rather than copied, so no stale libc is dragged forward, and
/usr/local, /opt and the non-bind-mounted parts of /workspace are copied
verbatim with tar --skip-old-files so they can never clobber a newer
base binary.
docker diff is not used: on a snapshot-derived container it reports only
changes since the last commit. Raw image-vs-image diffing is filtered
through dpkg ownership because it otherwise lies — 8,677 raw path
differences on a real project reduced to 2 genuinely user-authored
files, both loose /workspace-root files.
Crash safety. snapshot:latest keeps pointing at the old image until the
final commit, so any crash before it self-heals on next start. Later
crashes are caught by reconcile_project_statuses. The rollback pin is a
docker tag: 0.057s and 0 bytes. Rollback restores the system layer only
— volumes are never touched — and the UI says so rather than implying a
time machine.
Fixes an infinite recreation loop shipped with the MCP removal. docker
commit propagates labels to the image, so a container created from a
snapshot inherited its non-empty triple-c.mcp-fingerprint and the
one-shot shim recreated it again on every start, forever. Lineage labels
are now always written explicitly.
Documents the second, separate bug this uncovered: Dockerfile changes
under /home/claude never reach an existing project, migration or not,
because the volume masks them. Anything that must stay upgradable
belongs in /usr/local/bin or /opt, or must be seeded by entrypoint.sh.
145 Rust tests, 227 frontend tests, both builds clean.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Four features, plus a latent bug fix.
llama.cpp backend. Claude Code only ever speaks the Anthropic Messages
API — confirmed empirically by pointing it at a logging server, which
received POST /v1/messages?beta=true. llama-server implements that
natively (verified in its README, alongside --port default 8080), so
this is a plain base-URL backend with no translation shim, the same
shape as Ollama. Its --api-key defaults to none, so the auth token is a
placeholder Claude Code requires and llama-server ignores.
Model alias fix. ANTHROPIC_DEFAULT_HAIKU_MODEL is documented as "also
used for background functionality", and Triple-C set none of the alias
vars. So on every custom-endpoint backend, Claude Code resolved `haiku`
to an Anthropic model id and sent it to a local server that does not
have it — background features failed silently. All four
ANTHROPIC_DEFAULT_{OPUS,SONNET,HAIKU,FABLE}_MODEL vars are now pinned to
the backend's configured model, with an optional Haiku override, and
blanked for Anthropic and Bedrock so those keep Claude Code's defaults.
The deprecated ANTHROPIC_SMALL_FAST_MODEL is never emitted. Existing
Ollama and OpenAI-Compatible containers are recreated once so the new
env reaches them; the snapshot is preserved.
Model gateway. Optional LiteLLM sibling container, off by default,
mirroring stt.rs — this is what makes real OpenAI usable, since
api.openai.com has no /v1/messages. Pinned to v1.96.0 by tag and digest:
the 1.82.7/1.82.8 malware was PyPI-only and never affected the official
images, which is precisely why this builds FROM the image rather than
pip-installing, but 1.84.0 is still the floor for proxy CVEs (API-key
SQLi, Host-header auth bypass, MCP auth bypass). Binds 0.0.0.0 because
project containers consume it, and therefore always sets a master_key —
LiteLLM without one accepts any key. The provider key lives in the OS
keychain and is uploaded into a volume, never an image layer or label.
URL relay. A container-side xdg-open/BROWSER shim opens URLs in the
host's browser. Uses an OSC sequence to /dev/tty rather than a printed
sentinel, because the shim usually runs as a grandchild of a process
capturing its children's output. Degrades to printing the URL when no
terminal is attached, so scheduled tasks do not hang. Only http/https,
with control characters rejected before new URL() — which strips
newlines, so java\nscript: would otherwise parse as javascript:. Nothing
auto-opens; the user confirms. The web terminal shows a tap-to-open
banner instead, since that browser may be a phone across a tunnel.
Browser view. A Project Home tab that watches and takes over the browser
Claude drives with Playwright, using Playwright's own dashboard. Zero
image cost — Playwright stays user-installed. It does not reuse the auth
bridge's PortForward, which binds an unauthenticated port: correct for a
throwaway OAuth listener, wrong for mouse and keyboard control of a
browser in a passwordless-sudo container. Instead a token-gated loopback
proxy checks Host, then token or a forbidden-header origin signal,
before a byte reaches the container. Host ports are confined to
47820..=47827 so CSP frame-src can enumerate them rather than widening
to a wildcard, with a test asserting the two agree.
188 frontend tests, 107 Rust tests, both builds clean.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Project Home (DESIGN-REVIEW §B2): the project is promoted from a 280px
sidebar card to a first-class main-area view. ProjectCard.tsx (1,257
lines) is replaced by a select-only ProjectRow plus tabs for Overview,
Sessions, Automation, Config and Files. The PortMappings, FileManager
and ContainerProgress modals are absorbed rather than reimplemented.
Config gains a Saved/Saving/Failed indicator — save-on-blur failures
previously reached only console.error.
Tier-1 polish (DESIGN-REVIEW §A): new elevation, muted-accent, disabled
and focus-ring tokens; a global :focus-visible ring with every
focus:outline-none removed; filled buttons moved to --accent-emphasis
and white-on-success toggles retired, fixing three WCAG AA failures
(2.1:1, 2.5:1, 2.4:1); a shared Modal primitive with role="dialog",
focus trap and restore, adopted by all remaining modals; status
indicators that carry a glyph and word rather than colour alone.
Ctrl+Shift+W closes a tab, deliberately not Ctrl+W — that is readline's
kill-word, used constantly in the terminal this app is built around.
Auth Bridge: a general loopback-callback bridge so browser logins run
inside a container (aws sso login, Concourse fly login, claude login)
can complete against the host browser. Listeners are discovered from
/proc/net/tcp{,6} — ss/netstat/lsof are absent from the image — bound on
host 127.0.0.1 only, and tunnelled in over the Docker API via socat,
which keeps working on Docker Desktop where container IPs are not
routable. Falls back to [::1] because Node resolves localhost to IPv6
first, so claude login often binds ::1 alone. Opt-in per project.
This extracts create_attached_exec() and moves the existing terminal
session path onto it, so there is one attached-exec implementation
rather than two.
Shared auth token: `claude setup-token` is run in a container, the token
is stored in the OS keychain and injected as CLAUDE_CODE_OAUTH_TOKEN
into Anthropic-backend projects. Contrary to the initial design note,
setup-token uses an Anthropic-hosted redirect and blocks on a stdin
paste prompt rather than a loopback callback, so a stdin command is
required for the flow to complete.
The token is never logged, never returned to the frontend, and is
redacted from the streamed output with a stateful matcher that withholds
any tail that could still grow into a secret. Change detection uses a
random rotation id rather than a hash, since a hash in a docker-inspect
readable label would be an offline verification oracle.
Frontend 33 -> 51 tests; Rust 34 tests. Both builds clean.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Permission modes: replaces the binary full_permissions flag with a
PermissionMode enum (Plan/Default/AcceptEdits/Bypass). Flag mapping is
defined once in PermissionMode::cli_args() and used by the terminal, the
web terminal, and the scheduler:
Plan -> --permission-mode plan
Default -> (no flag)
AcceptEdits -> --permission-mode acceptEdits
Bypass -> --dangerously-skip-permissions
Choices verified against `claude --permission-mode` on 2.1.226.
full_permissions is retained and effective_permission_mode() falls back
to it, so existing projects.json needs no migration.
Bug fix: triple-c-task-runner ran `claude -p ... --dangerously-skip-
permissions` unconditionally, ignoring the project's setting entirely.
It now reads TRIPLE_C_PERMISSION_MODE, which is injected into the
container, added to the reserved env blocklist, propagated through the
entrypoint's cron env filter, and tracked by a new
triple-c.permission-mode label so a change forces recreation.
Introspection: new commands/inspect_commands.rs exposes read-only views
into the container over docker exec — Claude sessions (parsed from
~/.claude/projects/<cwd>/<uuid>.jsonl), installed capabilities (skills,
agents, commands, hooks, plugins, natively-configured MCP servers), and
the triple-c-scheduler task list, logs and notifications.
Task/session ids are validated against a strict allowlist and every
parameterized call runs as a bare argv vector via bollard, so no shell
is involved. Stopped containers return empty results rather than errors.
No UI yet; that lands with the Project Home view.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Completes the removal begun in the previous commit.
Backend: deletes models/mcp_server.rs, storage/mcp_store.rs and
commands/mcp_commands.rs, the McpStore on AppState, the four IPC
handlers, Project::enabled_mcp_servers, build_mcp_servers_json(),
compute_mcp_fingerprint(), the MCP_SERVERS_JSON env injection, the
mcp-fingerprint label, and the whole MCP container lifecycle.
create_container() and container_needs_recreation() lose their
mcp_servers/network_name parameters.
Container: entrypoint.sh no longer merges MCP_SERVERS_JSON into
~/.claude.json. MCP_SERVERS_JSON stays in the reserved env blocklist.
Security: the Docker socket is no longer auto-mounted for stdio+Docker
MCP servers — it now mounts only when allow_docker_access is set.
Migration: old containers were created with
network_mode=triple-c-net-<projectId> and refuse to start once that
network is gone. docker/network.rs becomes docker/legacy_cleanup.rs with
label-driven, best-effort removal of leftover MCP containers and the
per-project network, called on both delete and recreate.
container_needs_recreation() now forces a rebuild for any container
carrying a non-empty triple-c.mcp-fingerprint label or attached to a
triple-c-net-* network, moving it onto the default bridge. Both can be
dropped a release later.
Docs: drops the MCP sections from README/HOW-TO-USE/TECHNICAL and adds a
short note pointing at Claude Code's native `claude mcp` / `/mcp` /
.mcp.json instead.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
When Docker isn't detected on startup, surface a dialog offering a
one-click install (pkexec + get.docker.com on Linux, brew cask on
macOS, winget on Windows) with a graceful fallback to manual steps
and a link to official documentation. Install output streams back
to the UI via a tauri event.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds a mic button to the terminal UI that captures speech, transcribes
it via a Faster Whisper sidecar container, and injects the text into
the terminal input. Includes settings panel for model selection
(tiny/small/medium), port config, and container lifecycle management.
- stt-container/: Dockerfile + FastAPI server for Whisper transcription
- Rust backend: STT container management, transcribe_audio IPC command
- Frontend: useSTT hook, SttButton, SttSettings, WAV encoder
- CI: Gitea Actions workflow for multi-arch STT image builds
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Adds an axum HTTP+WebSocket server that runs alongside the Tauri app,
serving a standalone xterm.js-based terminal UI accessible from any
browser on the local network. Shares the existing ExecSessionManager
via Arc-wrapped stores, with token-based authentication and automatic
session cleanup on disconnect.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Help dialog now fetches HOW-TO-USE.md live from the gitea repo on open,
falling back to the compile-time embedded copy when offline. Content is
cached for the session. Removes the ~600-line hardcoded markdown constant
from HelpDialog.tsx in favor of a single source of truth.
Adds a Table of Contents with anchor links for quick navigation and a new
troubleshooting entry for the "Failed to install Anthropic marketplace"
error with the jq fix. Markdown renderer updated to support anchor links
and header id attributes.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
SSO login was broken in containers due to three issues: the sso_session
indirection format not being resolved by Claude Code's AWS SDK, SSO
detection only checking sso_start_url (missing sso_session), and the
OAuth callback port not being accessible from inside the container.
This fix runs SSO login on the host OS (where the browser and ports work
natively) by having the container emit a marker that the Tauri app
detects in terminal output, triggering host-side `aws sso login`. The
entrypoint also inlines sso_session properties into profile sections and
injects awsAuthRefresh into Claude Code config for mid-session refresh.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Adds two new features for running project containers:
1. Bash Shell Tab: A "Shell" button on running projects opens a plain
bash -l session instead of Claude Code, useful for direct container
inspection, package installation, and debugging. Tab labels show
"(bash)" suffix to distinguish from Claude sessions.
2. File Manager: A "Files" button opens a modal file browser for
navigating container directories, downloading files to the host,
and uploading files from the host. Supports breadcrumb navigation
and works with any path including those outside mounted projects.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add Model Context Protocol (MCP) server configuration support. Users can
define MCP servers globally (new sidebar tab) and enable them per-project.
Enabled servers are injected into containers as MCP_SERVERS_JSON env var
and merged into ~/.claude.json by the entrypoint.
Backend: McpServer model, McpStore (JSON + atomic writes), 4 CRUD commands,
container injection with fingerprint-based recreation detection.
Frontend: MCP sidebar tab, McpPanel/McpServerCard components, useMcpServers
hook, per-project MCP checkboxes in ProjectCard config.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Feature 1 - Update Detection: Query Gitea releases API on startup (3s
delay) and every 24h, compare patch versions by platform, show pulsing
"Update" button in TopBar with dialog for release notes/downloads.
Settings: auto-check toggle, manual check, dismiss per-version.
Feature 2 - Multi-Folder Projects: Replace single `path` with
`paths: Vec<ProjectPath>` (host_path + mount_name). Each folder mounts
to `/workspace/{mount_name}`. Auto-migrate old single-path JSON on load.
Container recreation via paths-fingerprint label. AddProjectDialog and
ProjectCard support add/remove/edit of multiple folders.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Tauri v2 desktop app (React/TypeScript + Rust) for managing
containerized Claude Code environments. Includes Gitea Actions
workflow for building and pushing the sandbox container image,
and a BUILDING.md guide for manual app builds on Linux and Windows.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
shadow-testandClaude Opus 5