Skip to content

Repository files navigation

Reasonledger

A notation for transferring reasoning between AI systems that leaves an auditable record of the transfer.

Reasonledger encodes model-to-model messages as named blocks, each carrying an explicit instruction for how the receiver should process the enclosed content — activate this as a premise, enforce this as a constraint, verify this derivation step by step. Across a multi-hop chain of agents, those blocks and a three-way epistemic status accumulate into an inspectable record of what each model claimed, what it verified, and what it accepted on trust. That record is the ledger.

Reasonledger is a payload format. It rides inside a transport such as MCP or A2A; it does not replace one. It is not a blockchain and involves no tokens, chain, or consensus — "ledger" refers to the audit trail the format produces, nothing more.

Why it exists

Multi-agent systems are easy to run and hard to inspect. When a planner hands work to an executor, and that output feeds a third agent, the reasoning that produced the final answer is scattered across opaque messages. Plain prose loses structure at each hop; plain JSON carries data but no record of what was checked versus assumed. Reasonledger keeps the structure and adds the record: which agent asserted what, which steps were verified, which were trusted, and where a claim's status changed along the way.

The intended use is governance and observability for agentic pipelines — a reasoning ledger you can read after the fact to see how a conclusion was reached and how much of it was actually verified.

What's in the format

  • Content blocks@FORMAL, @LOGIC, @DERIVE, @TAXONOMY, @ANCHOR — typed reasoning with per-block processing instructions.
  • Exchange blocks@STATE, @VERIFY, @OUTPUT — state handoff, per-step verification reporting, response shaping.
  • Transfer-protocol blocks@PROVENANCE, @TRANSFER_REPORT, @DERIVATION_TRACE, @SCOPE, @CONFIDENCE, @TRUST_TRANSFORMATION — chain-of-custody, per-hop integrity, and trust calibration for multi-hop chains.
  • Three-way epistemic statusconfirmed / open / denied, preserved across model boundaries rather than collapsed.

Full grammar and processing rules are in SPEC.md.

A small example

@FORMAL [kepler_third]
T² = (4π² / GM) · a³

@DERIVE [circular_velocity]
(1) For a circular orbit, gravity supplies the centripetal force
(2) G·M·m / r² = m·v² / r
(3) From (2): v² = G·M / r
→→ v = sqrt(G·M / r)

@TAXONOMY [orbit_properties]
bound_orbit: confirmed
exact_eccentricity: open
parabolic_escape: denied

After a transfer hop, the receiver's record carries the audit trail:

@PROVENANCE [chain]
  hop_1: model_A → model_B | loaded+full_verify
@VERIFY [§circular_velocity]
(1): holds
(2): holds
(3): holds
→→: holds

Reference validator

A small, dependency-free Python validator checks a document against the v1.0 grammar (recognized block types, the required basis attribute on @LOGIC, valid status markers, reference syntax):

python reference/validator.py examples/orbital.rl

It exits non-zero and prints line-referenced errors on any violation. The validator is reference code under Apache 2.0.

Status

The format has been exercised in author-run experiments across five model architectures, where all five adopted the notation in their own output and preserved the three-way epistemic status across multi-hop transfer. These are author-run results; independent replication is invited and has not yet been performed. Reasonledger makes no claim about model cognition.

License

This repository is dual-licensed:

  • Specification and documentation (SPEC.md, grammar, prose docs) — Creative Commons Attribution 4.0 International (CC BY 4.0), see LICENSE-CC-BY-4.0.
  • Code (reference checker, evaluation harness, scripts) — Apache License 2.0, see LICENSE.

Reasonledger is an open, royalty-free standard, freely implementable by anyone, including in commercial products.

Citing

See CITATION.cff. Short form:

Aiello, M. P. (2026). Reasonledger: Specification v1.0. Zenodo. https://doi.org/10.5281/zenodo.19874728

Lineage

Reasonledger was developed and first disclosed as AI-Native Notation (ANN). The earlier name is retained in the provenance record for citation and priority; see NOTICE and the specification's "Version lineage" section.

About

Open specification and reference checker for auditing reasoning transfer in multi-agent LLM systems. Companion to "When the Auditor Cries Wolf": an LLM auditor's false-positive rate on clean traces moves from 60/60 to 0/60 on prompt content alone.

Topics

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages