open-knowledge-studio (oks)
File-based knowledge engineering CLI for Claude Code and coding agents.
oks is the command-line core of Open Knowledge Studio:
a file-based knowledge base that turns raw material into a recallable, self-decaying wiki.
Install
pipx install "git+https://github.com/open-agent-power/open-knowledge-studio.git@main#subdirectory=cli" && pipx ensurepath
We recommend pipx because modern Linux (Ubuntu 24.04+) and macOS Homebrew Pythons are
PEP 668 externally-managed, so a bare pip install fails. If your mirror lags behind
PyPI, add --pip-args="-i https://pypi.org/simple".
Optional multimodal ingest is included in the package, while its heavy dependencies (PDF / audio / video / formula extraction) remain opt-in:
oks capability install watch
oks capability install document
oks capability install pdf
Add --yes to install a listed capability after reviewing its dependencies.
What you get
- 6+1-factor recall engine — token overlap, substring, topic trace, type boost, review penalty, memory curve, plus an optional goal boost that lifts on-scope pages.
- Dreaming cycle — distill raw materials into draft proposals; humans review and promote them to the wiki.
- Decay system — memory-curve scoring with type-specific λ and hot/warm/cold/evictable tiers.
oksCLI — search, recall, offline evaluation, execution traces, wiki CRUD, drafts, distill, lint, status, metrics.
The CLI core is dependency-light and calls no external network APIs; agents and humans orchestrate the pipeline around it.
Quick start
pipx install "git+https://github.com/open-agent-power/open-knowledge-studio.git@main#subdirectory=cli" && pipx ensurepath # 1. install the CLI
oks init my-knowledge-base # 2. scaffold an instance (skills + buckets)
cd my-knowledge-base
oks status # 3. use it
oks recall "git branch" --knowledge-only
oks recall "deployment" --knowledge-only --type strategy --format json
oks recall "git branch" --goal none --format json --explain
oks eval recall eval/datasets/team-v1.yaml --output eval/runs/baseline.json
oks trace start memory-goal --run-id demo-001
Use --goal active (default) to merge active goals, --goal <slug> for one
reproducible goal, or --goal none for a no-goal baseline. --explain exposes
score components without changing ranking.
Machine-readable output uses recall-response/v1 for oks recall and
recall-hit/v1 for individual hits. --type filtering happens before ranking
and --limit; --knowledge-only drops the episodic path.
oks eval is offline and read-only. oks trace writes append-only execution
events under raw/executions/; generated Wiki/Skill proposals stay under
drafts/proposals/ until a human explicitly promotes them.
oks init materializes the shareable layer (Claude Code skills, templates, schema,
settings) and a git-tracked memory instance. No repo clone required.
Documentation
- Design docs: https://open-agent-power.github.io/open-knowledge-studio/
- Source & issues: https://github.com/open-agent-power/open-knowledge-studio
License
MIT
1 maintainer
Metadata
1 maintainer
Release files for open-knowledge-studio 0.6.21
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | |
|---|---|---|---|---|
| open_knowledge_studio-0.6.21.tar.gz | 288.3 kB | With provenance | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| open_knowledge_studio-0.6.21-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 599.5 kB
Release files / open_knowledge_studio-0.6.21.tar.gz
| Download URL | open_knowledge_studio-0.6.21.tar.gz |
|---|---|
| Size | 288.3 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
c69b424687bc79f9e5aefc53221ab7575614d07c58a4367edf33a0420c4f0c0a
|
|
BLAKE2b-256 checksum How to use checksums |
2809c44f29aeca518c4d1d3a1f884e397a91c98bb98d4a3f4590549bf15a54e5
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
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 Aug 31, 2026.
Transparency logRelease files / open_knowledge_studio-0.6.21-py3-none-any.whl
| Download URL | open_knowledge_studio-0.6.21-py3-none-any.whl |
|---|---|
| Size | 311.2 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
973e5dc13fd947130f1c3c2a296e6e00f63166edc8abbfab0ef0bf5fafcf71d9
|
|
BLAKE2b-256 checksum How to use checksums |
f7b6cebe533dca0e821f79ff6ca71ef98393d10e5b35bae83b7c3f0d3939fe2c
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
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 Aug 31, 2026.
Transparency logSecurity for open-knowledge-studio 0.6.21
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 2 files in this release have been uploaded with provenance metadata.
Attestation for open_knowledge_studio-0.6.21.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 Aug 31, 2026.
Transparency logA 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.
