Skip to main content

Create isolated feature worktrees with allocated runtime environments.

Project description

new-feature

new-feature aims to eliminate friction for creating new features. In a single command, it creates isolated git worktrees, allocates conflict-free runtime values, runs project-defined setup and teardown commands (such as creating and deleting a feature database), and launches the configured coding agent in the new worktree.

Install

uv tool install new-feature
new-feature setup

new-feature setup launches the configured coding agent (Codex by default) in the current repository. The agent inspects the project, proposes a repository-specific configuration, asks about unresolved choices such as the optional Codex hook, and waits for approval before editing. Run it again to review and improve an existing integration.

Releases

Changing [project].version in pyproject.toml and merging that change to main publishes the distributions to PyPI, tags the merged commit as v<VERSION>, and creates a GitHub Release. Keep the project version in uv.lock synchronized with pyproject.toml. The release workflow can also be run manually to recover from a partial release.

Usage

# Ask an agent to configure or improve new-feature for this repository.
new-feature setup
# Create a feature worktree and launch the configured agent.
new-feature my-feature
# Create a feature worktree without launching another agent.
new-feature my-feature --no-agent
# Merges the worktree into the main branch
new-feature merge my-feature
# Run configured teardown and remove the feature worktree.
new-feature teardown my-feature
# Teardown a worktree even if it has uncommitted work.
new-feature teardown my-feature --force
# Inspect managed features and diagnose stale state.
new-feature list
new-feature doctor
new-feature doctor --repair

Codex hook

Install the Codex hook in the current repository:

new-feature install-codex-hook

This writes .codex/hooks.json; Codex loads the guard only for this trusted repository.

Project Config

Add config to the target repo's new-feature.toml:

target_branch = "main"
default_agent = "codex"
agents = { codex = ["codex"], claude = ["claude"] }
push = false
setup = ["uv sync"]
pre_merge = ["uv run pytest"]
post_merge = ["uv run pytest"]
teardown = []

[env]
WEB_PORT = { allocate = "port", min = 3000, max = 3999 }
API_PORT = { allocate = "port", min = 4000, max = 4999 }
DATABASE_NAME = { allocate = "name", prefix = "myapp", max_length = 63 }
CACHE_DIR = { allocate = "path", base = ".new-feature/cache" }

All settings remain optional. If both new-feature.toml and pyproject.toml exist, new-feature.toml takes precedence. For projects that prefer to keep tool configuration in pyproject.toml, place the same settings under [tool.new-feature] and use [tool.new-feature.env] instead of [env].

default_agent is the configured agent name used when --agent is omitted. Codex and Claude are built in; agents adds or overrides named commands and fixed arguments. new-feature appends its generated feature prompt as the final argument, so agents that require a prompt flag can be configured directly:

default_agent = "custom"
agents = { custom = ["custom-agent", "--prompt"] }

Use a configured agent for one invocation, or pass an executable command directly:

new-feature my-feature --agent claude
new-feature my-feature --agent "fooagent --baz-flag"
new-feature setup --agent claude

When the value exactly matches a key in agents, that configured command is used. Otherwise, new-feature parses the value as an executable command without invoking a shell.

The built-in create and setup prompts can be overridden with create_prompt and setup_prompt in TOML, or for one invocation with --prompt.

setup runs after worktree creation; teardown runs before worktree removal.

Supported env entries:

  • { value = "literal" }
  • { allocate = "port", min = 3000, max = 3999 }
  • { allocate = "integer", min = 1, max = 15 }
  • { allocate = "name", prefix = "myapp", max_length = 63 }
  • { allocate = "slug", prefix = "myapp" }
  • { allocate = "path", base = ".new-feature/cache" }

Lifecycle

new-feature my-feature creates branch my-feature and worktree .worktrees/my-feature, reserves env values in .new-feature/manifest.toml, runs setup, and launches the configured agent in the worktree. It automatically adds .new-feature/ and .worktrees/ to .gitignore. If setup fails, it runs a forced teardown so the partial worktree, branch, and manifest entry do not linger.

new-feature list shows each managed feature and its current Git/worktree state. new-feature doctor reports stale manifest entries, dirty worktrees, unmerged branches, and configuration drift. doctor --repair removes only manifest entries whose worktree and branch are both already gone.

new-feature merge my-feature runs pre-merge checks in the feature worktree, starts a no-commit merge into the target branch, runs post-merge checks on the merged target checkout, commits the merge only if those checks pass, and pushes only when push = true.

new-feature teardown my-feature runs the configured teardown commands before removing the worktree, deleting the branch, and removing the manifest entry. If the worktree has uncommitted changes or the branch has commits that are not in the target branch, pass --force to abandon them deliberately.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

new_feature-1.0.1.tar.gz (22.5 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

new_feature-1.0.1-py3-none-any.whl (28.4 kB view details)

Uploaded Python 3

File details

Details for the file new_feature-1.0.1.tar.gz.

File metadata

  • Download URL: new_feature-1.0.1.tar.gz
  • Upload date:
  • Size: 22.5 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.12

File hashes

Hashes for new_feature-1.0.1.tar.gz
Algorithm Hash digest
SHA256 9269e4ae6dd74fe63a8f54125d79248b30fd455fb1ab70cc196c389e83308978
MD5 3e56472a388ef52a30aff7b58fdbf256
BLAKE2b-256 9ceb02fc7388c5202e323f9b2c2ab5db1ed80a3c6e6a166c5f9167ab11c87d17

See more details on using hashes here.

Provenance

The following attestation bundles were made for new_feature-1.0.1.tar.gz:

Publisher: publish.yml on crypdick/new-feature

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file new_feature-1.0.1-py3-none-any.whl.

File metadata

  • Download URL: new_feature-1.0.1-py3-none-any.whl
  • Upload date:
  • Size: 28.4 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.12

File hashes

Hashes for new_feature-1.0.1-py3-none-any.whl
Algorithm Hash digest
SHA256 7ac7a655115ae4a89d501854462ee780738a2caea7c756efaee290e8b03f7156
MD5 d49a5c72444a17f9d65ff0894250479a
BLAKE2b-256 5292de501ab74f29696cb4f49e3d8b1e51b99e4edded9dfb2b4704113a128487

See more details on using hashes here.

Provenance

The following attestation bundles were made for new_feature-1.0.1-py3-none-any.whl:

Publisher: publish.yml on crypdick/new-feature

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page