This release adds 3 notable features for engineering teams evaluating rollout.
✓ No known CVEs patched in this version
Topics
+14 more
Affected surfaces
Summary
AI summaryIntroduces ClaimVerifier to independently verify agent actions and enforce trust decay for forged claims.
Changes in this release
| Type | Severity | Summary | CVE |
|---|---|---|---|
| Feature | Medium |
Adds ClaimVerifier subsystem for independent action verification. Adds ClaimVerifier subsystem for independent action verification. Source: llm_adapter@2026-06-08 Confidence: high |
— |
| Feature | Medium |
Introduces ExecutionReceipt HMAC‑signed receipt for exec() and writeFile(). Introduces ExecutionReceipt HMAC‑signed receipt for exec() and writeFile(). Source: llm_adapter@2026-06-08 Confidence: high |
— |
| Feature | Medium |
Extends ComplianceMonitor with UNSUPPORTED_CLAIM and UNDISCLOSED_ACTION violation types. Extends ComplianceMonitor with UNSUPPORTED_CLAIM and UNDISCLOSED_ACTION violation types. Source: llm_adapter@2026-06-08 Confidence: high |
— |
| Feature | Medium |
Adds AuthGuardian methods recordClaimViolation(), resetClaimViolations(), getClaimViolationCount(), and getTrustLevel(). Adds AuthGuardian methods recordClaimViolation(), resetClaimViolations(), getClaimViolationCount(), and getTrustLevel(). Source: llm_adapter@2026-06-08 Confidence: high |
— |
| Feature | Medium |
Updates AgentResult.metadata to include receipts field. Updates AgentResult.metadata to include receipts field. Source: llm_adapter@2026-06-08 Confidence: high |
— |
| Feature | Medium |
Updates index.ts to export new ClaimVerifier types. Updates index.ts to export new ClaimVerifier types. Source: llm_adapter@2026-06-08 Confidence: high |
— |
| Feature | Low |
Adds 50 new tests for ClaimVerifier across three phases. Adds 50 new tests for ClaimVerifier across three phases. Source: llm_adapter@2026-06-08 Confidence: high |
— |
| Refactor | Low |
Modifies lib/agent-runtime.ts to emit ExecutionReceipt after exec/write. Modifies lib/agent-runtime.ts to emit ExecutionReceipt after exec/write. Source: llm_adapter@2026-06-08 Confidence: high |
— |
| Refactor | Low |
Modifies security.ts to define ExecutionReceipt interface and receipt utilities. Modifies security.ts to define ExecutionReceipt interface and receipt utilities. Source: llm_adapter@2026-06-08 Confidence: high |
— |
| Refactor | Low |
Updates THREAT_MODEL.md with Tier 1 scope ceiling and Tier 2 hardening path. Updates THREAT_MODEL.md with Tier 1 scope ceiling and Tier 2 hardening path. Source: llm_adapter@2026-06-08 Confidence: high |
— |
Full changelog
ClaimVerifier — Tier 1 Agent Honesty / Lie Detector
Network-AI v5.10.0 ships ClaimVerifier, a new subsystem that lets the orchestrator independently verify whether an agent's claimed actions actually happened — catching fabricated results, exaggerated outcomes, and undisclosed side-effects.
How it works
AgentRuntime now issues an HMAC-signed ExecutionReceipt for every exec() and writeFile() call, committing to { agentId, action, target, exitCode, outputHash }. The runtime — not the agent — is the sole outcome authority. Tampering with any field (including exitCode or outputHash) invalidates the signature.
ClaimVerifier.verify(manifests, agentId, windowMs) reconciles agent-declared ActionManifest[] against the in-memory RuntimeAuditEntry log:
- UNSUPPORTED_CLAIM — agent declared an action that has no matching runtime witness (forged or imagined)
- UNDISCLOSED_ACTION — runtime witnessed an action the agent did not declare
Both violation types surface through ComplianceMonitor and can trigger AuthGuardian trust decay.
Trust decay
AuthGuardian.recordClaimViolation() increments a per-agent counter on each UNSUPPORTED_CLAIM. After N consecutive violations (default 3) trust drops by 0.1. Below 0.4 trust, the agent is forced into ApprovalGate supervised execution.
esetClaimViolations() resets the counter on a corroborated turn.
New exports
ypescript import { ClaimVerifier, ActionManifest, VerificationOutcome, VerificationResult } from 'network-ai' import { ExecutionReceipt } from 'network-ai'
Scope (Tier 1)
ClaimVerifier operates on AgentRuntime-mediated actions only. Unmediated BYOC adapter network calls, per-session audit log boundaries, and the interpretive gap between "ran" and "succeeded semantically" are documented in THREAT_MODEL.md section 8 as Tier 2 hardening candidates (capability broker + process isolation + egress-deny).
Changes
- New: lib/claim-verifier.ts — ClaimVerifier, ActionManifest, VerificationOutcome, VerificationResult, ClaimVerifierOptions
- New: est-claim-verifier.ts — 50 tests (Phase 1: receipt, Phase 2: reconciliation, Phase 3: trust decay)
- Modified: security.ts — ExecutionReceipt interface; SecureTokenManager.generateReceipt() / alidateReceipt()
- Modified: lib/agent-runtime.ts — ShellResult.receipt?, FileResult.receipt?, receipt emission after exec/write
- Modified: lib/auth-guardian.ts —
ecordClaimViolation(),
esetClaimViolations(), getClaimViolationCount(), getTrustLevel() - Modified: lib/compliance-monitor.ts — ViolationType extended with UNSUPPORTED_CLAIM, UNDISCLOSED_ACTION
- Modified: ypes/agent-adapter.d.ts — AgentResult.metadata.receipts?
- Modified: index.ts — exports for all new types
- Modified: THREAT_MODEL.md — section 8 Tier 1 scope ceiling + Tier 2 hardening path
- Docs: README, SECURITY, .github/SECURITY, SKILL.md, CONTRIBUTING, AUDIT_LOG_SCHEMA, CHANGELOG, ARCHITECTURE — all updated for v5.10.0
Test suite
3211 tests across 32 suites (50 new + 3161 existing), 0 failures.
Full changelog
See CHANGELOG.md.
Weekly OSS security release digest.
The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.
No spam, unsubscribe anytime.
Share this release
About Jovancoding/Network-AI
Multi-agent orchestration MCP server with race-condition-safe shared blackboard. 20+ MCP tools: blackboard read/write, agent spawn/stop, FSM transitions, budget tracking, token management, and audit log query. `npx network-ai-server --port 3001`.
Related context
Beta — feedback welcome: [email protected]