← Back to Products

Runtime monitoring and tamper-evident audit.
The complete post-deployment layer.

Evidence-AIDR (AI Detection and Response) watches your AI agents in production — detecting attacks, intercepting dangerous actions, and sealing every runtime event into a cryptographically signed evidence bundle any auditor can verify offline. No raw data exposed. No vendor dependency.

Talk to us

Runtime monitoring

Runtime threat detection
for agents in production.

Evidence-AIDR sits in your production environment watching every agent action in real time — not just logging, but actively detecting and intercepting dangerous behavior before it causes damage.

Attack Detection

Prompt injection, policy bypass, and privilege abuse flagged in live agent traffic as they happen.

Prompt injection · policy bypass · anomaly detection

Dangerous Action Interception

High-risk actions — shell calls, secret access, unauthorized writes — scored and blocked before execution.

Pre-execution scoring · auto-block · signed verdict

Incident Replay & Forensics

Every agent action timestamped and signed. Replay any incident from the cryptographic record — no agent self-reporting required.

Deterministic replay · tamper-evident · independent forensics

tracestone.io

Cryptographic properties

Every runtime event sealed
with eight verifiable properties.

Every evidence bundle is checked against P1–P8. The open-source verifier reproduces all eight locally — offline, from a single binary, in under 5 minutes.

P1
Chain Integrity

Merkle hash chain — any row insertion or deletion is detectable by re-hashing

P2
Attribution

Every log row provably linked to a specific tenant and session via sub-chain isolation

P3
Signature

ed25519 signature from the gateway keypair on every bundle manifest

P4
Anti-tamper

Manifest + row-level checksums prevent silent modification after signing

P5
Redaction Proofs

Removed PII (Personally Identifiable Information) rows leave Merkle proofs — auditor can confirm shape without seeing content

P6
Privacy

PII declared and redacted; CMK (Customer Managed Key) envelope encryption; pseudonymization at gateway

P7
TEE Attestation

Verifier runs in Nitro Enclave; PCR0 (Platform Configuration Register 0) bound to OSS commit hash — even cloud provider can't read prompts

P8
PRE Authorization

Proxy re-encryption auditor handover; time-limited grant expires and revokes instantly

Open source · Apache 2.0

The verifier is open source.
The trust is yours.

The OSS verifier is the trust root of the entire system. Any auditor, regulator, or engineer can download it, point it at an evidence bundle, and reproduce the result — offline, no account, no NDA.

tracestone.io
Evidence bundles with PASS and FAIL verdict

Evidence bundles — Ed25519-signed packages your auditors download and re-verify offline with a single command. PASS or FAIL, independently. 

# Install (darwin / linux / windows)
$ curl -sSL https://install.tracestone.io | bash

# Verify the bundled demo — zero config required
$ tracestone verify --demo
✓ P1 chain ok      1,034 rows, prev_hash chain intact
✓ P3 sig   ok      ed25519 verified against gateway.pub
✓ P5 redact ok     merkle proofs for 12 redacted rows
✓ P6 privacy ok    PII declared + CMK envelope
✓ P7 attest ok     Enclave PCR0 in allowlist
✓ P8 authz  ok     pre-grants.json sig verified

PASS  proof_hash sha256:bb162c…

# Point at your own bundle
$ tracestone verify audit-bundle-2026-05.zip

tracestone-ai/verifier

Apache 2.0 · P1–P8 Go CLI + library · single static binary · cross-platform (darwin / linux / windows)

tracestone-ai/llm-gateway-audit-skill

LLM Gateway vulnerability corpus + audit prompts · portable skill bundle · Apache 2.0

Privacy architecture

Your raw data never
leaves your boundary.

Tracestone is designed so we provably cannot read your prompts. Pseudonymization and encryption happen inside your VPC before any data is transmitted. Our SaaS holds only ciphertext and signed evidence.

Customer VPC
Raw data never leaves in plaintext
  • AI Gateway / Agent Emits audit.jsonl rows as events occur
  • Pseudonymization HMAC + per-tenant salt · PII fields replaced before leaving VPC
  • CMK Envelope Encryption AWS KMS · Azure Vault · HSM (Hardware Security Module) · revoke key = data instantly inaccessible
  • P6 PII Scanner email · phone · ID · credit card · Luhn + GB-11643 validation
Tracestone SaaS
Holds only ciphertext & signed evidence
  • Multi-tenant Ingestion P2 attribution · sub-chain isolation · manifest signature validation
  • TEE Confidential Verify OSS verifier runs inside AWS Nitro Enclave / Azure CC · attestation bound to OSS commit hash · even AWS cannot read prompts
  • Proxy Re-Encryption (PRE) Time-limited auditor access grant · one-click revocation · zero plaintext throughout
  • Differential Privacy Aggregation CISO dashboard stats without exposing individual records
Auditor · Regulator
Independent · offline · no NDA
  • OSS Verifier Single static binary · cross-platform · reproduces proof_hash · checks P1–P8
  • PRE Recipient Customer issues grant → auditor decrypts authorized fields · grant expires / revokes instantly
  • Zero-Knowledge Compliance Proof Prove compliance to regulator · original data stays in enterprise boundary · zkVM (RISC Zero / SP1)
  • Auto-generated Compliance Reports SOC 2 · ISO 42001 · EU AI Act · HIPAA · NIST AI RMF

Compliance coverage

Evidence mapped to the
frameworks auditors use.

SOC 2 Type II

Automated evidence pack generation with P1–P8 crosswalk to CC6–CC9 common criteria.

ISO 42001

AI management system audit trail — provenance, traceability, and accountability requirements.

EU AI Act

High-risk AI system provenance and technical documentation requirements (Articles 11–17).

HIPAA

PHI handling with CMK envelope encryption + P6 PII redaction + audit log integrity.

NIST AI RMF

Govern · Map · Measure · Manage alignment — evidence bundles map to each function.

MAS TRM (SG)

Financial sector AI governance for APAC — evidence chain satisfies MAS Technology Risk Management guidelines.

tracestone.io
Compliance map showing EU AI Act, ISO 42001, NIST AI RMF, SOC 2 controls with PASS/FAIL status

AI agents running in production?

Evidence-AIDR monitors your agents in real time, intercepts dangerous actions, and builds the audit trail your compliance team needs — without exposing any raw data. Start with the free OSS verifier or talk to us directly.

Talk to us