salluru.dev
operational

All systems active

Labs / Verification & Trust
Launching

tracemarket

Prove what your agent did — to an auditor, an insurer, or a regulator.

An open protocol for verified agentic execution traces — canonicalize, scrub, hash, publish.

At a Glance
open
trace format
hash-chain
integrity
DP
privacy scrub
signed
attestations
The Problem

Every regulated team deploying agents now has to produce tamper-evident, attributable execution records — but “we logged it” is not “we can prove what the agent did and that the log wasn't doctored.” Today's agent logs are framework-specific blobs with PII baked in, no verification that the recorded steps produced the recorded outcome, and no way to share an incident trace without leaking secrets.

Key Insight

Verification doesn't require re-running the task — it requires checking internal consistency: that recorded preconditions, actions, and tool returns are mutually consistent and hash-chain-intact. The unit of value is the verified, attestable trace itself, so the buyer of verification is the trace's own owner — no data-pooling prisoner's dilemma.

How It Works

Canonical format

One open, framework-agnostic schema for agent traces — the wire format others emit and verify against.

Consistency verifier

A post-hoc state-diff check that the trace's steps are internally consistent and the hash chain is intact.

Privacy scrub

A structure-preserving scrub that survives an LLM re-identification adversary while keeping semantic content.

Attestation registry

Signed, regulator-recognizable attestations with revocation and chain-of-custody.

VerificationProvenanceCompliance