CITATION.cff

cff-version: 1.2.0
message: "If you use LATTICE in academic work, please cite it as below."
type: software
title: "LATTICE: Ledgered Agent Traces for Transparent, Inspectable Collaborative Execution"
abstract: >-
  LATTICE is the accountability layer that sits underneath any
  agent framework, or raw Python functions. Every agent decision
  becomes a content-addressed (SHA-256), cryptographically signed
  (Ed25519) claim in a DAG you can trace backward from any
  conclusion to raw evidence. Effective confidence (min-
  propagation) prevents weak evidence from hiding behind strong
  conclusions; the revocation waterfall propagates compromise
  status to dependents; cycle detection enforces DAG acyclicity
  at insertion time; the audit pass flags unsupported claims,
  low confidence, broken references, and inflated confidence.
  Includes a CLI, a local FastAPI + D3.js dashboard, and a
  zero-friction monitor decorator for instrumenting existing
  Python functions without behavioural change.
authors:
  - given-names: "Ali Murtaza"
    family-names: "Bhutto"
    orcid: "https://orcid.org/0009-0007-2787-943X"
    affiliation: "SZABIST University (MSc Cybersecurity, 2026)"
repository-code: "https://github.com/thunderstornX/lattice"
url: "https://github.com/thunderstornX/lattice"
license: MIT
version: "1.2.2"
date-released: "2026-05-22"
doi: "10.5281/zenodo.20341935"
identifiers:
  - type: doi
    value: "10.5281/zenodo.20341934"
    description: "Concept DOI (always resolves to the latest version)"
  - type: doi
    value: "10.5281/zenodo.20341935"
    description: "Version DOI (v1.2.2)"
keywords:
  - multi-agent systems
  - accountability
  - audit log
  - content addressing
  - Ed25519
  - SHA-256
  - claim DAG
  - effective confidence
  - revocation waterfall
  - OSINT
  - AI safety
  - explainable AI