A simple coding harness with planner-implementor-validator orchestration
Project description
kunchi
Multi-agent coding harness for Python: planner → implementor → validators, with output contracts and retry loops.
Install from PyPI:
pip install kunchi
CLI
Initialize a project, then run a goal in your repository:
cd /path/to/your/repo
kunchi init --template cursor # or antigravity, fake, minimal
kunchi init --template antigravity --force # replace an existing profile
export CURSOR_API_KEY=... # required for the cursor template
kunchi run "Your goal"
kunchi run "Your goal" --monitor # compact progress on stderr
kunchi run "Your goal" --trace # verbose debug trace on stdout
kunchi plans
kunchi resume PLAN_ID
Use kunchi init --template fake to try the flow without Cursor or other external agent CLIs.
Try the bundled demo (fake agents, no external dependencies):
kunchi-demo "Build a demo feature"
Python API
Fake agents (default) — no external dependencies:
from kunchi import Harness, default_harness_config
harness = Harness(default_harness_config())
report = await harness.run("Build feature")
Cursor CLI — requires the agent binary on PATH:
from kunchi.harness import Harness, default_cursor_harness_config
harness = Harness(default_cursor_harness_config(workspace_path="/path/to/repo"))
report = await harness.run("Build feature")
Profiles — bundle output contracts and validators:
from kunchi.harness import Harness, harness_config_from_profile
harness = Harness(harness_config_from_profile("cursor", "/path/to/repo"))
Load a custom profile from YAML or JSON:
from kunchi.harness import Harness, harness_config_from_profile_file
harness = Harness(harness_config_from_profile_file("my-profile.yaml", "/path/to/repo"))
Persist and resume plans:
from kunchi import Harness, default_harness_config
from kunchi.state import FilePlanStore
harness = Harness(default_harness_config(), plan_store=FilePlanStore(".kunchi/plans"))
report = await harness.run("Build feature")
report = await harness.resume(report.plan_id)
summaries = await harness.list_plans()
Documentation
- Design — architecture, agents, validators, and extension points
- Generic CLI provider plan — config-driven agents (
cli: generic) - Roadmap — milestones and planned work
- Contributing — development setup and manual testing
License
MIT — see LICENSE.
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
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file kunchi-0.1.3.tar.gz.
File metadata
- Download URL: kunchi-0.1.3.tar.gz
- Upload date:
- Size: 43.8 MB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.2.0 CPython/3.12.3
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
30d795cda7dc8dfa5f6d25285a65d0dcce651034c9565e1e744c695b3f61f713
|
|
| MD5 |
a4ae4de3ed6c33afd0b4cdd2e099a0e4
|
|
| BLAKE2b-256 |
d0613e80003abb152cc612ed800ff25deab178c81f4d0d59857b221092f07b0d
|
File details
Details for the file kunchi-0.1.3-py3-none-any.whl.
File metadata
- Download URL: kunchi-0.1.3-py3-none-any.whl
- Upload date:
- Size: 82.5 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.2.0 CPython/3.12.3
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
8e31ea1b85285af1e70f2003bd27ddff02ebe1cde7239fdcbbab2c6618eb2185
|
|
| MD5 |
12dbe93ecd1a54a08b89b208aba8e6bc
|
|
| BLAKE2b-256 |
6bf43f6031e02c746457a08dfe42c0c075df8c744b8af4cbda3036b439d0051e
|