Skip to main content

TrustGraph — Jac/Jaseci OSP agentic walker for trustandverify knowledge verification

Project description

trustgraph-jac

TrustGraph Jac/Jaseci OSP walker for agentic knowledge verification.

Built on top of trustandverify — the Python package that implements all confidence algebra, search backends, storage, and export. This package provides the Jac/Jaseci graph layer: OSP nodes, edges, byLLM functions, and the agentic walker.

Enhanced from the Velric Miami Hackathon 2026 implementation. Fully additive — hackathon usage works unchanged.


Install

pip install trustgraph-jac

Requires API keys:

export GEMINI_API_KEY=...
export TAVILY_API_KEY=...

Usage

Same as hackathon (unchanged)

jac run trustgraph.jac
jac run trustgraph.jac "Is nuclear energy safer than solar?"
jac run trustgraph.jac "Is coffee healthy?" --claims 5

New in 0.2.0

# Multi-backend search (Tavily + Brave + Bing, deduped)
jac run trustgraph.jac "Is remote work productive?" --search multi

# Export all formats
jac run trustgraph.jac "Is coffee healthy?" --format all

# Persist to SQLite
jac run trustgraph.jac "Is coffee healthy?" --storage sqlite --db reports.db

# Full options
jac run trustgraph.jac "Is nuclear energy safer than solar?" \
  --claims 5 \
  --search multi \
  --storage sqlite \
  --db reports.db \
  --format all

What's in this package

File Purpose
trustgraph.jac Main walker — additive to hackathon, new flags for backends
bridge/confidence.py Drop-in replacement for hackathon bridge/confidence.py — delegates to trustandverify.scoring
tools/search.py Drop-in replacement for hackathon tools/search.py — delegates to trustandverify.search

What moved to trustandverify

All confidence algebra, search backends, storage, and export logic now lives in the trustandverify Python package. The Jac layer focuses on what Jac is good at: graph structure, byLLM functions, and the agentic walker pattern.


Hackathon compatibility

The bridge modules preserve the exact import signatures from the hackathon:

import from bridge.confidence {
    scalar_to_opinion, flip_opinion, fuse_evidence,
    apply_trust_discount, detect_conflicts_within_claim,
    opinion_summary, build_jsonld_claim
}
import from tools.search { web_search }

These still work — they now delegate to trustandverify instead of having inline implementations.


New node fields (0.2.0)

The Claim node now stores explicit opinion components for graph queries:

claim.opinion_belief      # float
claim.opinion_disbelief   # float  
claim.opinion_uncertainty # float

Previously only claim.confidence (projected probability) and claim.status (verdict string) were stored.


Part of the TrustGraph project. Built at the Velric Miami Hackathon 2026.

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

trustgraph_jac-0.2.0.tar.gz (11.6 kB view details)

Uploaded Source

Built Distribution

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

trustgraph_jac-0.2.0-py3-none-any.whl (12.9 kB view details)

Uploaded Python 3

File details

Details for the file trustgraph_jac-0.2.0.tar.gz.

File metadata

  • Download URL: trustgraph_jac-0.2.0.tar.gz
  • Upload date:
  • Size: 11.6 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: poetry/2.3.2 CPython/3.12.2 Windows/11

File hashes

Hashes for trustgraph_jac-0.2.0.tar.gz
Algorithm Hash digest
SHA256 0b5bd17b651a27a7db1d2bafa4325869dc98908da0e3d0c10e95990544e2c62c
MD5 0c3a38c0e1a811d1cb2fdbfb053c5763
BLAKE2b-256 35df0a05a4d80a87a5c90faea48a14fd717689248c9da9ecec91a2a68e834ba1

See more details on using hashes here.

File details

Details for the file trustgraph_jac-0.2.0-py3-none-any.whl.

File metadata

  • Download URL: trustgraph_jac-0.2.0-py3-none-any.whl
  • Upload date:
  • Size: 12.9 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: poetry/2.3.2 CPython/3.12.2 Windows/11

File hashes

Hashes for trustgraph_jac-0.2.0-py3-none-any.whl
Algorithm Hash digest
SHA256 f495aa88505c7db6780ca04b21cda75c2347d8471a7c5b0247758d69922a9dac
MD5 7a71374ab43c47dd804fc661a440448e
BLAKE2b-256 ef200b13222baa0d666738cb27d16e437647773df0931949d452bbdd16cbc247

See more details on using hashes here.

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