Skip to main content

agent6

A coding agent with sandboxed commands, resumable sessions, and editable workflows.

Terminal UI showing a coding task, command approval, verification, and changes
agent6 tui
terminal interface
Web UI showing sessions, tool details, and sandbox configuration
agent6 web
browser interface for desktop or phone
  • Runs commands in a Linux sandbox; you set file and network access.
  • Routes each role to a hosted, local or subscription model.
  • Resumes or forks a saved session from an earlier turn.
  • Runs your verification command and marks an unverified tree.
  • Reviews the diff with a model panel; blocking findings stop a merge into the checkout branch.
  • Runs in a worktree, or as ranked parallel attempts in clones. When agent6 controls git, it merges the winner into the parent's chain, not into the checkout branch. A top-level run may dispatch children itself when harness.parallel.tool is on.
  • Defines workflows as TOML state machines with recovery and replay.
  • Follows AGENTS.md and SKILL.md packs; MCP servers are opt-in.
  • No telemetry unless you export it, and no automatic updates.

Also available as a streaming CLI and through ACP editor integration.

Installation

Sandboxing supports Linux on x86_64 and aarch64; macOS runs without it, and Windows users should use WSL.

Install from PyPI with uv or pipx, then connect a provider and select a model:

uv tool install agent6      # or pipx install agent6
agent6 connect              # configure a provider
agent6 model default        # choose the required default model

Use Anthropic or any OpenAI-compatible endpoint, including OpenAI, OpenRouter, Ollama, vLLM, llama.cpp, and LM Studio. A provider entry can also use the Responses API (providers.<name>.backend = "responses") or sign in at a configured OAuth issuer. For subscriptions, replace agent6 connect with agent6 connect chatgpt or agent6 connect claude; the latter uses your installed, signed-in Claude Code binary. agent6 completions installs tab completion for your shell.

Usage

A run edits your current checkout. Checkpoint commits go on a separate agent6/<id> branch; checkpointing leaves your current branch's history and index in place. For a separate checkout, use agent6 run --worktree "your task", which starts from committed HEAD without copying uncommitted changes or untracked dependencies.

cd your-repo
agent6 check                # check configuration and sandbox support
agent6 run "your task here"

Commands ask for approval by default. After the task finishes, enter another instruction or type /exit.

Inspect the latest session, then merge its recorded changes when ready:

agent6 sessions show        # status and verification result
agent6 sessions diff        # recorded changes
agent6 sessions merge       # merge into the run's starting branch

Other commands:

Command Purpose
agent6 plan "task" plan a change; agent6 run --from <plan-id> carries it out
agent6 ask "question" ask about the repository
agent6 attach <id> follow a live session
agent6 steer <id> "instruction" direct the next steps
agent6 sessions findings <id> list a session's review findings
agent6 sessions receipt <id> print the recorded tests, execution environments, review outcome, the children it dispatched and how to revert
agent6 stop <id> --after-turn stop after the current turn
agent6 resume <id> continue an interrupted run
agent6 fork <id> --at-turn 7 fork a coding run from an earlier turn into a separate worktree
agent6 telemetry export export session journals as OpenTelemetry signals or a Prometheus textfile
agent6 pack copy your config, secrets, sessions and memory into the workspace before a sync
agent6 unpack list what the pack would install into the new home, ask, then move it in

To give a finished run more work, use agent6 resume <id> --steer "next task".

Configuration

agent6 config show prints effective settings and where each value came from. Changes are global unless you add --repo for a repository override, for example:

agent6 config set --repo harness.verify.command '["uv", "run", "pytest"]'

To measure one run, pass --metric 'minimize -- python3 bench.py' to agent6 run.

The worker, planner, reviewer, and helper inherit each unset model, effort, and temperature from the required models.default. Session records identify who made each model call by its actor. Budgets limit spending, tokens for unpriced calls, and supported subscription usage.

Security

agent6 confines only the model's commands: in a jail on your host under full and partial isolation, or inside the dev container you select; the agent6 process stays trusted, holds your provider credentials, and connects to your configured providers. The repository stays at its real path, and the jail sees your system toolchains and ~/.local/bin read-only.

sandbox.isolation names the command mode; auto, the default, picks the strongest of full, partial and none the host allows and warns when it falls back:

  • full: commands cannot read your provider keys or the secrets file, .git is read-only to them (sandbox.protect_git = "auto", git.control = "agent6"), and they get a private network (sandbox.network = "auto").
  • partial: commands cannot read your keys or the secrets file unless a path you grant contains it, but they share agent6's network and can write .git.
  • none: no confinement.
  • container: commands run inside the dev container that mounts your workspace, found by its labels or named in [container]; the runtime is the boundary, and agent6 adds no confinement there.

Commands ask for approval by default (sandbox.run_commands = "ask"). agent6's own git never pushes, rewrites history or runs reset --hard, and no config key turns that on. MCP is off by default, and locally spawned servers use the command sandbox unless explicitly configured otherwise. agent6 check sandbox shows the effective file and command boundaries, network and approval settings, and live probe results.

Running agent6 itself in Docker or Podman

  • Docker gives partial by default and allows full with both --security-opt seccomp=unconfined and --security-opt apparmor=unconfined.
  • Rootless Podman allows full with no flags (rootful Podman).

On Ubuntu, full under Docker also needs the host to allow the jail launcher to use unprivileged user namespaces (Installation).

Metadata

Release files for agent6 0.0.44

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for agent6 0.0.44
File Size Uploaded Provenance Actions
agent6-0.0.44.tar.gz 2.1 MB With provenance Details for agent6-0.0.44.tar.gz

Built distributions (wheels)

Table of built distributions (wheels) for agent6 0.0.44
File Interpreter ABI Platform Actions
agent6-0.0.44-py3-none-manylinux_2_17_x86_64.musllinux_1_2_x86_64.whl Python 3 none Linux musl 1.2+ x86-64, Linux glibc 2.17+ x86-64 Details for agent6-0.0.44-py3-none-manylinux_2_17_x86_64.musllinux_1_2_x86_64.whl
agent6-0.0.44-py3-none-manylinux_2_17_aarch64.musllinux_1_2_aarch64.whl Python 3 none Linux glibc 2.17+ ARM64, Linux musl 1.2+ ARM64 Details for agent6-0.0.44-py3-none-manylinux_2_17_aarch64.musllinux_1_2_aarch64.whl

Total release size: 7.5 MB

Release files / agent6-0.0.44.tar.gz

Download URL agent6-0.0.44.tar.gz
Size 2.1 MB
Tags Source
SHA-256 checksum
How to use checksums
ad7718467e5839a25fef6e328a412f8c927a73677c4e684d169a0971431dc099
BLAKE2b-256 checksum
How to use checksums
c89118f149e5c2c3d1c310ac8ec6881f5d95d24da8ee3a7184f6c3883e1b37a5
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.16

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Nov 20, 2026.

Transparency log

Release files / agent6-0.0.44-py3-none-manylinux_2_17_x86_64.musllinux_1_2_x86_64.whl

Download URL agent6-0.0.44-py3-none-manylinux_2_17_x86_64.musllinux_1_2_x86_64.whl
Size 2.8 MB
Tags Linux glibc 2.17+ x86-64 Linux musl 1.2+ x86-64 Python 3
SHA-256 checksum
How to use checksums
dc6a6e3e747367f468d006bf9cd2be717af69bd6838d0a28f41d10931876c0f2
BLAKE2b-256 checksum
How to use checksums
f076207080f808e2bcf73c5db15892d9e2e1955bfa69c52b3844000a78cb9c7c
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.16

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Nov 20, 2026.

Transparency log

Release files / agent6-0.0.44-py3-none-manylinux_2_17_aarch64.musllinux_1_2_aarch64.whl

Download URL agent6-0.0.44-py3-none-manylinux_2_17_aarch64.musllinux_1_2_aarch64.whl
Size 2.7 MB
Tags Linux glibc 2.17+ ARM64 Linux musl 1.2+ ARM64 Python 3
SHA-256 checksum
How to use checksums
f5641f2313af8974180300c6460c5380181082af44b87d7a6f6b01e87ae46bc2
BLAKE2b-256 checksum
How to use checksums
7b3dcc428ccfc2f21297903fa905941295ffd643ebabbec502f5e830ca69ba09
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.16

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Nov 20, 2026.

Transparency log

Security for agent6 0.0.44

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

Full provenance

All 3 files in this release have been uploaded with provenance metadata.

Select from 3 files

Attestation for agent6-0.0.44.tar.gz

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Nov 20, 2026.

Transparency log

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Release history Release notifications | RSS feed

This release

0.0.44 This release

3 release files Full provenance

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page