claude-bottle

Run multiple Claude Code agents on your own machine, each scoped to its own secrets, skills, and egress allowlist.

Four probes against a real bottle, end-to-end: pipelock forwards a clean HTTPS GET to an allowlisted host, blocks a GET to a non-allowlisted host, blocks a POST whose body carries a credential pattern; git-gate rejects a push containing a leaked key. Run it yourself with bash scripts/demo.sh.

Why "claude-bottle"?

Each container is a bottle; Claude is the genie inside. The genie's powers are exactly what the manifest grants it — a specific set of skills, a specific set of secrets, and a specific set of hosts it can reach — nothing more. You uncork one bottle per agent (./cli.py start <agent>), many bottles run in parallel, and each is scoped to its task. When the session ends the bottle is destroyed and the genie does not persist.

Goals

• Scope each agent to the minimum credentials and network egress its task actually needs
• Run multiple agents in parallel, isolated from each other
• Keep code, credentials, and agent activity on infrastructure I control — no third-party agent runtime

Security model

Each agent runs in its own bottle: its own container, its own internal Docker network, and its own pipelock sidecar. Bottles don't share state, don't talk to each other, and only get the env vars, skills, SSH identities, and egress hosts the manifest grants them — nothing more. Any one agent only has the access it needs to do its job.

The bottle limits both what an agent can see and where it can send it. Each bottle gets only the secrets and SSH identities the manifest grants it — a Gitea token but not a GitHub token, a deploy key but not a personal SSH key — so even a compromised or misbehaving agent only handles credentials it was already trusted with for its job. Egress flows through pipelock, which constrains where those credentials can travel: an agent with a Gitea token can reach gitea.dideric.is, not arbitrary attacker-controlled hosts. The same constraint blocks DNS-over-HTTPS as an exfil channel — a DoH resolver like cloudflare-dns.com would have to be on the allowlist for the agent to reach it at all. The container itself adds a layer between the agent and the host, but the v1 design leans more on secret minimization and egress allowlisting than on the container as a hardened boundary. On Linux hosts where gVisor is registered with Docker, claude-bottle auto-detects it and launches every bottle under runsc for a userspace syscall barrier — no manifest configuration required. The broader v2 discussion lives in docs/research/stronger-isolation-alternatives.md.

The egress proxy and OAuth-token handling below are the load-bearing pieces of v1.

Architecture

A bottle is the agent container plus up to three per-protocol egress sidecars on a per-agent Docker --internal network. The agent has no default route off-box; its only way out is through the pipelock sidecar (for HTTP/HTTPS), the ssh-gate sidecar (for SSH), or the git-gate sidecar (for git operations against declared upstreams). Each sidecar also sits on an egress network that does have internet access, so the agent's traffic always passes through a container that enforces the manifest before it leaves the host.

                            host  ( ./cli.py )
                                  │
                          starts  │  stops
                                  ▼
   ┌─────────────────────────── bottle ──────────────────────────┐
   │                                                             │
   │   ┌──────────────────┐                                      │
   │   │ agent image      │   HTTPS_PROXY    ┌────────────────┐  │  HTTPS to
   │   │ (claude-code,    │ ───────────────► │ pipelock image │──┼──►  allowlisted
   │   │  built locally)  │                  │ (TLS bump, DLP,│  │     hosts
   │   │                  │                  │  allowlist)    │  │
   │   │ skills, env,     │                  └────────────────┘  │
   │   │ ~/.gitconfig     │                                      │
   │   │                  │   git ops        ┌────────────────┐  │  SSH (push/
   │   │                  │ ───────────────► │ git-gate image │──┼──►  fetch) to
   │   │                  │                  │ (gitleaks +    │  │     bottle.git
   │   │                  │                  │  git daemon)   │  │     upstreams
   │   └──────────────────┘                  └────────────────┘  │
   │                                                             │
   │   agent on internal network (no default route);             │
   │   sidecars also attached to an egress network.              │
   └─────────────────────────────────────────────────────────────┘

• agent image — built from the repo Dockerfile (node:22-slim base) on first run; runs claude with the manifest-granted skills, env vars, and ~/.gitconfig (the latter for the git-gate's insteadOf rules when bottle.git is set).
• pipelock image — per-agent sidecar. Terminates the agent's outbound HTTP/HTTPS, enforces the resolved allowlist, runs DLP scanning. Design in docs/prds/0001-per-agent-egress-proxy-via-pipelock.md and docs/prds/0006-pipelock-tls-interception.md.
• git-gate image — per-agent sidecar built on zricethezav/gitleaks (alpine + gitleaks + git-daemon + openssh-client). Runs git daemon over git:// as a bidirectional mirror of each declared upstream. A pre-receive hook gitleaks-scans incoming refs and forwards clean refs to the real upstream over SSH; an access-hook runs git fetch origin --prune against the upstream before every upload-pack so an agent fetch returns whatever the upstream has now (fail-closed if unreachable). The agent's ~/.gitconfig rewrites the real URL to the gate via insteadOf, so push, fetch, clone, and pull all route through. The agent never sees the upstream credential. If the upstream's hostname isn't resolvable from the gate container (e.g. a Tailscale-only host whose public DNS points elsewhere), pin its IP via ExtraHosts: { "<hostname>": "<ip>" } on the bottle.git entry — the gate's /etc/hosts gets the override while the agent's insteadOf rewrite still keys off the original hostname. Brought up only when bottle.git has entries. Design in docs/prds/0008-git-gate.md.

When the agent exits, cli.py tears down every sidecar that was brought up and the two networks; nothing about a bottle persists between runs.

Quickstart

Requires Docker on the host and a long-lived Claude Code OAuth token in your shell env.

./cli.py start <agent>   # builds the image on first run, drops you into claude

The container is removed automatically when the session ends. If the script is killed with SIGKILL the exit trap won't fire and the container may be left running; remove it with docker rm -f <container-name>.

Manifest

Agents and the bottles they run in are declared in claude-bottle.json in your project root or $HOME (both files merge if present, with project entries overriding home entries on key conflict).

{
  "bottles": {
    "gitea-dev": {
      "env": {
        "GITEA_TOKEN":     "?paste your Gitea API token",
        "GITHUB_TOKEN":    "${GH_PAT}",
        "GIT_AUTHOR_NAME": "didericis"
      },

      "git": [
        {
          "Name":         "claude-bottle",
          "Upstream":     "ssh://git@gitea.dideric.is:30009/didericis/claude-bottle.git",
          "IdentityFile": "/Users/didericis/.ssh/id_ed25519_gitea",
          "KnownHostKey": "ssh-ed25519 AAAA..."
        }
      ],

      // Egress is forced through a per-agent
      // [pipelock](https://github.com/luckyPipewrench/pipelock) sidecar
      // on a Docker `--internal` network — without the proxy the agent
      // has no route off-box. The effective allowlist is the union of
      // baked-in defaults (api.anthropic.com, claude.ai, ...) and the
      // hostnames listed here. Pipelock also runs DLP scanning and
      // detects URL-embedded high-entropy secrets. The resolved
      // allowlist is shown in the y/N preflight before launch.
      "egress": {
        "allowlist": [
          "github.com",
          "registry.npmjs.org",
          "pypi.org"
        ]
      }
    }
  },

  "agents": {
    "gitea-helper": {
      "bottle": "gitea-dev",
      "skills": ["init-prd"],
      "prompt": "You help maintain Gitea-hosted projects."
    }
  }
}

Comments are illustrative; the file itself must be valid JSON. See claude-bottle.example.json for a working starting point. Pipelock's design lives in docs/prds/0001-per-agent-egress-proxy-via-pipelock.md and the rationale in docs/research/pipelock-assessment.md.

Auth: OAuth token, not API key

claude-bottle authenticates claude inside the container with the same Pro/Max subscription you already use on the host, via a long-lived OAuth token. No ANTHROPIC_API_KEY is needed.

Why a token instead of mounting ~/.claude.json: on macOS, Claude Code stores OAuth credentials in the encrypted Keychain, not in ~/.claude.json. Mounting that file into a Linux container does not carry the credentials with it. Linux hosts keep credentials in ~/.claude/.credentials.json, but to keep the launcher portable claude-bottle uses the env-var path on every host.

One-time setup on the host:

claude setup-token   # browser login, prints a ~1-year OAuth token

Stash the token in your shell env (e.g. ~/.zshrc or a secret manager) as CLAUDE_BOTTLE_OAUTH_TOKEN:

export CLAUDE_BOTTLE_OAUTH_TOKEN="<token>"

cli.py automatically forwards it to every container as CLAUDE_CODE_OAUTH_TOKEN via docker run -e — no manifest wiring required, and the value is never written to disk or placed on argv.

Inside the container, claude picks up CLAUDE_CODE_OAUTH_TOKEN and authenticates against your subscription. Caveats: the token is bound to your subscription tier (Pro/Max/Team/Enterprise), it does not work with claude --bare (which only reads ANTHROPIC_API_KEY), and if it leaks, regenerate via claude setup-token again. Reference: https://code.claude.com/docs/en/authentication.

Trademarks

claude-bottle is an independent project and is not affiliated with, endorsed by, or sponsored by Anthropic, PBC. "Claude" and "Claude Code" are trademarks of Anthropic, PBC; the project name uses "claude" descriptively to indicate that the tool runs Claude Code inside a sandbox.

License

Copyright 2026 Eric Bauerfeld

Licensed under the Apache License, Version 2.0. See LICENSE for the full text.