Amni-Prism
Shared knowledge atlas built on PRISM-TEX (.ptex). Every fact gets a unique nonce — contribute knowledge, it deduplicates automatically by content hash, and the atlas grows. NDJSON manifests are append-only and git-merge-friendly, so multiple contributors never collide.
Status: alpha (0.1.x). API may shift before 1.0.
Install
pip install amni-prism
For local development:
git clone https://github.com/Amnibro/Amni-Prism
cd Amni-Prism
pip install -e .
Quick Start
import prism
prism.contribute_text('./codex', 'The speed of light is 299792458 m/s', domain='physics')
result = prism.query_text('./codex', 'The speed of light is 299792458 m/s')
hits = prism.search_keyword('./codex', 'speed of light')
CLI
prism contribute "The speed of light is 299792458 m/s" -d physics
prism search "speed of light"
prism propose "Water boils at 100C at sea level" -d physics --confidence 0.8
prism pending
prism verify
prism scrape --file article.txt --source "https://en.wikipedia.org/wiki/Light"
prism stats
How It Works
- Nonces — Every word/line/block gets a unique nonce in finite-field space. Same content → same nonce → automatic dedup.
- PRISM-TEX Storage — Knowledge stored as nonce-addressed maps. Vocab in NLX format, hierarchical tiers in HNA format.
- Content-Hash Dedup — SHA-256 over normalized content ensures no duplicates across contributors merging different codexes.
- Two-Tier Verification — Small models propose (~50 tokens), large models verify (~20 tokens). ~70 tokens per verified fact.
- NDJSON Manifests — Append-only JSON-lines manifests are git-merge-friendly. No merge conflicts when two contributors add facts to the same domain.
Domains
25 built-in domains: code, math, science, history, language, music, art, philosophy, psychology, economics, politics, law, medicine, engineering, technology, education, sports, entertainment, food, travel, nature, religion, mythology, literature, general.
Why use this
- Knowledge that survives model swaps. The atlas is a pile of facts addressed by content, not a fine-tune of a specific model. Swap your LLM tomorrow, keep your atlas.
- Federation without coordination. Two contributors building independent codexes can
prism mergetheir work and get a deduplicated union, with no central server. - Built for the Adam runtime. Amni-Prism is the federated layer of Amni-Ai (Adam). The
[federated]extra in Adam pulls this in directly.
Source Allowlist
Scraping is gated to license-compatible sources (Wikipedia, arXiv, RFC, Python docs, MDN, HuggingFace). See prism.scrape.ALLOWED_SOURCES.
Contributing
Fork, clone, pip install -e ., add facts via the CLI or Python API, and open a PR with your codex export. NDJSON manifests merge cleanly under git.
License
MIT — see LICENSE.
Metadata
Release files for amni-prism 0.1.2
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 | Actions |
|---|---|---|---|---|
| amni_prism-0.1.2.tar.gz | 22.1 kB | Without provenance | Details for amni_prism-0.1.2.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| amni_prism-0.1.2-py3-none-any.whl | Python 3 | none | any | Details for amni_prism-0.1.2-py3-none-any.whl |
Total release size: 43.6 kB
Release files / amni_prism-0.1.2.tar.gz
| Download URL | amni_prism-0.1.2.tar.gz |
|---|---|
| Size | 22.1 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
1f72845d0ec2db19d7e248dd96880c1f4fbe53798b1da2c15119e3e7ac6cb6ca
|
|
BLAKE2b-256 checksum How to use checksums |
72bf59452a5f219b185c9fe1a759a4845ea75fa0ee2fa916c67e3c7f24e722ed
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.12.10
|
Release files / amni_prism-0.1.2-py3-none-any.whl
| Download URL | amni_prism-0.1.2-py3-none-any.whl |
|---|---|
| Size | 21.5 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
161a3ed2b439ee0a4ac060421d679745a18fb81e720655bcedbc7fd9cb962d27
|
|
BLAKE2b-256 checksum How to use checksums |
0f2a2e175e64b34783e2c404f45a2a17a2257304cb6e9d8f8acab17b48bfe904
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.12.10
|
Security for amni-prism 0.1.2
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.
No provenance found
None of the files in this release were uploaded with provenance metadata.
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.