Release list

URL Intelligence Agent v1.3.0 — Semantic Conflict Intelligence Latest

Latest

vpicciuolo released this 08 Sep 15:07

v1.3.0

98f785a

🧠 URL Intelligence Agent v1.3.0

Semantic Conflict Intelligence

v1.3.0 significantly improves how URL Intelligence Agent understands contradictory evidence.

Until now, different wording across Meta tags, Open Graph, Twitter Cards, structured data and page content could sometimes be interpreted too aggressively as a conflict.

The new semantic conflict engine distinguishes normal wording differences from genuine factual disagreement and logical contradiction.

🔥 What's new

🧠 Semantic Conflict Classification

The provenance resolver can now classify relationships including:

semantic_equivalent
wording_variation
compatible_range
precision_difference
numeric_drift
temporal_drift
factual_disagreement
logical_contradiction
value_conflict

This provides substantially more precise evidence analysis than simple string comparison.

✍️ Wording Variation Detection

Different descriptions no longer automatically become hard conflicts simply because the wording is different.

For example:

Meta description

Open-source URL intelligence agent for web research and verification.

Open Graph

Evidence-first URL intelligence for research, provenance and web analysis.

These may now be classified as compatible wording variation rather than contradictory evidence.

🔎 Semantic Equivalence

The engine now normalizes and compares meaningful terms instead of relying only on exact string equality.

It evaluates:

Unicode normalization
punctuation
case
meaningful token overlap
semantic-token containment
factual anchors
numbers
currencies
URLs
explicit negation

⚠️ Factual Disagreement

When two claims contain materially different factual anchors, the engine can identify them as a real factual disagreement.

Examples include:

different prices
incompatible numeric claims
different URLs
identity differences
incompatible structured facts

These remain visible with their original provenance and evidence sources.

🚨 Logical Contradiction Detection

v1.3.0 introduces explicit contradiction detection for statements that share substantially the same subject but reverse their meaning.

Example:

Service is available worldwide.

vs.

Service is not available worldwide.

This is classified as a logical contradiction rather than ordinary wording variation.

Negation-aware analysis includes common patterns across multiple languages.

🔬 Evidence Inspector Upgrade

The hosted Evidence Inspector now makes these relationships easier to understand.

Users can inspect:

competing values
source layer
source property
representation
normalized value
confidence
observation timestamp
selected resolver value
conflict severity
semantic relationship
resolution explanation

Compatible semantic variations are now visible instead of being hidden behind the final resolved value.

🌐 Improved Conflict Explorer

Repeated conflicts across multiple crawled pages are grouped into reusable conflict patterns.

Instead of displaying:

value_conflict, value_conflict, value_conflict

the interface shows the actual competing evidence and explains why each relationship was classified.

⚡ Improved Web Experience

The live web application also includes an improved analysis workflow.

After clicking Run Analysis:

an analysis overlay opens immediately
investigation stages are displayed
elapsed analysis time is shown
provenance and external evidence are processed
when analysis finishes, the user is automatically taken directly to the result

No fake progress percentage is shown.

🧪 Regression Testing

New regression tests cover:

Meta / Open Graph / Twitter description variation
semantic wording differences
explicit negation
logical contradiction
strict factual disagreement
existing provenance behavior
numeric compatibility
representation drift
claim verification
SSRF protections

Current semantic/provenance suite:

17 / 17 tests passing

CI successfully validated the release on:

Node.js 18
Node.js 20
Node.js 22

🔐 Existing Security Boundaries Preserved

The semantic intelligence upgrade does not weaken existing security protections.

The project continues to include:

SSRF protections
private/reserved address blocking
DNS rebinding protections
redirect validation
evidence isolation
restricted hosted browser boundaries
Remote MCP abuse protections

🤗 Live Demo

The v1.3.0 runtime is deployed on Hugging Face:

https://huggingface.co/spaces/vpicciuolo/url-intelligence-agent

🔌 Remote MCP

https://vpicciuolo-url-intelligence-agent.hf.space/mcp

📊 Benchmark Dataset

https://huggingface.co/datasets/vpicciuolo/url-intelligence-benchmark

💻 Source Code

https://github.com/vpicciuolo/url-intelligence-agent

Created by Vincenzo Picciuolo
HRN Innovation Technologies Ltd

Developed inside the HORNO Network ecosystem.

MIT Licensed.

Assets 2

Release title: v1.2.0 — Claim Provenance & Temporal Consistency Engine

vpicciuolo released this 08 Sep 09:42

v1.2.0

9cc9925

URL Intelligence Agent v1.2.0

Claim Provenance & Temporal Consistency Engine

v1.2.0 is a major verification and web-intelligence upgrade for URL Intelligence Agent.

The agent now goes beyond extracting a single value from a page. It preserves field-level evidence, tracks exactly where each value came from, compares multiple page representations, detects conflicts and drift, reasons about approximate values and numeric ranges, and explains why one claim was selected over another.

🔎 Field-Level Provenance

Evidence can now include:

source representation
extraction layer
property and locator
raw value
normalized value
timestamps
confidence signals
evidence hashes

Supported evidence layers include:

Source HTML
Rendered DOM
Open Graph
Meta tags
Twitter Cards
JSON-LD
Microdata
RDFa
Visible content
HTTP headers
Runtime API evidence where permitted

🧠 Claim Resolution

New claim states include:

consensus
compatible_variation
drift
conflict
insufficient_evidence

The engine understands the difference between exact values, approximations, ranges and lower/upper bounds.

Example:

Open Graph: 80,000+
Rendered page: 100,502

Instead of incorrectly reporting a contradiction, v1.2 can determine:

Logical conflict No
Precision difference Yes
Representation drift Yes
Freshness divergence Yes
Stale metadata suspected Yes
Preferred value 100502

⏱ Temporal Consistency

Added support for:

representation drift detection
freshness divergence
stale metadata detection
temporal claim history
provenance-aware snapshots
ETag and Last-Modified evidence
conditional HTTP requests

🔐 Evidence Integrity

Evidence can include SHA-256 integrity hashes so downstream systems can verify exactly which representation produced a result.

A W3C PROV-shaped export is also available.

🛠 New Actions

New capabilities include:

inspect_provenance
verify_claim

Existing investigation, SEO, structured-data, comparison, monitoring and knowledge-export actions now also benefit from provenance intelligence.

🔌 MCP Upgrade

v1.2.0 supports:

MCP 2026-07-28
MCP 2025-11-25
MCP 2025-06-18
MCP 2025-03-26

Also included:

modern stateless MCP operation
legacy session compatibility
strict action-specific schemas
structured output schemas
MCP Tasks support
stronger hosted anti-abuse protections

🌐 Evidence Inspector

The live Hugging Face application now includes an Evidence Inspector for exploring:

observations
resolved claims
conflicts
drift
suspected stale metadata
representations
evidence layers
confidence
resolution explanations

📊 Benchmark

The v1.2.0 production release passed the current core benchmark:

55 / 55 benchmark cases
120 / 120 assertions
107 / 107 deterministic assertions
18 / 18 security checks
6 / 6 full-agent completion checks
100% current core benchmark score

A dedicated provenance regression track has also been added.

🛡 Security

This release also improves:

SSRF protection boundaries
browser request restrictions
private-address checks
download restrictions
Remote MCP rate-limit handling
charset-aware decoding
evidence isolation

Unrestricted browser rendering remains intentionally disabled on the public Hugging Face deployment because browser networking represents a separate SSRF and egress trust boundary.

🚀 Live

Hugging Face Space:
https://huggingface.co/spaces/vpicciuolo/url-intelligence-agent

Benchmark Dataset:
https://huggingface.co/datasets/vpicciuolo/url-intelligence-benchmark

GitHub:
https://github.com/vpicciuolo/url-intelligence-agent

Created by Vincenzo Picciuolo / HRN Innovation Technologies Ltd

Developed inside the HORNO Network ecosystem.

Licensed under the MIT License.

Assets 2

URL Intelligence Agent v1.1.0 - Network Security Hardening

vpicciuolo released this 08 Sep 04:30

v1.1.0

d277837

🔐 URL Intelligence Agent v1.1.0

Major network-security hardening for arbitrary URL analysis.

v1.1.0 strengthens the URL Intelligence Agent network boundary with connect-time DNS validation and expanded SSRF protection while preserving normal public-web compatibility.

🔥 What's new

DNS rebinding protection

The core safeFetch() pipeline now validates DNS not only before a request but also at the actual outbound socket connection.

This closes the DNS rebinding / TOCTOU gap where a hostname could theoretically resolve to a public IP during validation and later resolve to a private destination during connection.

Stronger SSRF protection

The network layer now uses CIDR-aware IPv4 and IPv6 classification.

Blocked destinations include:

loopback addresses
private networks
link-local addresses
cloud metadata ranges
reserved ranges
IPv4-mapped IPv6
NAT64 translation ranges
IPv6 transition/tunnel ranges
multicast and other non-public destinations

Mixed DNS responses containing both public and private addresses are rejected.

Redirect protection

Every redirect destination continues to be independently validated before it is followed.

Dedicated guarded network dispatcher

Untrusted URL collection now uses a dedicated Undici dispatcher with guarded DNS resolution.

The protection is scoped to URL Intelligence Agent network collection and does not modify Node.js global networking behavior.

🧪 Testing

v1.1.0 adds regression coverage for:

IPv4 SSRF destinations
IPv6 SSRF destinations
mapped and translated addresses
mixed public/private DNS responses
reserved address classes

Tested successfully on:

Node.js 18
Node.js 20
Node.js 22

📚 Documentation

This release adds and updates:

SECURITY.md
docs/NETWORK_SECURITY.md
docs/ARCHITECTURE.md
docs/DEPLOYMENT.md
docs/ACTIONS.md
CHANGELOG.md

⚙️ Runtime changes

Version bumped to 1.1.0
Minimum Node.js version: 18.17+
Added undici 6.x as a runtime dependency
User-agent versioning is now synchronized with the project version

🤗 Hugging Face

The updated v1.1.0 runtime has also been deployed to the official Hugging Face Space.

Live Demo:
https://huggingface.co/spaces/vpicciuolo/url-intelligence-agent

Remote MCP:
https://vpicciuolo-url-intelligence-agent.hf.space/mcp

⚠️ Renderer security boundary

The core HTTP collector is protected by the new connect-time DNS guard.

Optional Playwright/Chromium rendering operates through a separate browser networking boundary and should be deployed with infrastructure-level egress restrictions when enabled.

The public Hugging Face Space keeps optional browser rendering disabled by default.

Created by Vincenzo Picciuolo
HRN Innovation Technologies Ltd
Part of the HORNO Network ecosystem

GitHub:
https://github.com/vpicciuolo/url-intelligence-agent

Assets 2

URL Intelligence Agent v1.0.0 — Unified Intelligence Release

vpicciuolo released this 31 Aug 17:40

v1.0.0

b16e218

🧠 URL Intelligence Agent v1.0.0

URL in. Identity, evidence and intelligence out.

This is the first official public release of URL Intelligence Agent — an evidence-first, open-source web intelligence agent built for developers, AI agents, marketplaces, directories, research systems, monitoring and automation.

The technology is already used as part of the wider HORNO ecosystem and is now available publicly for developers to inspect, extend, self-host and integrate.

🔥 What it does

Give the agent a public URL and it can investigate what that URL represents, collect supporting evidence, discover connected public information and return structured intelligence with confidence and provenance.

Core capabilities

Deep website crawling and site mapping
robots.txt and sitemap discovery
URL safety and SSRF-aware fetching
Metadata and Open Graph extraction
JSON-LD and structured-data analysis
Entity resolution
Evidence and confidence scoring
Social profile discovery
Public contact discovery
People and team extraction
Technology fingerprinting
Brand intelligence
SEO and discoverability audits
Security-header posture analysis
Trust and transparency signals
DNS, mail and TLS intelligence
Entity relationship graphs
Competitor and comparison intelligence
API / OpenAPI / GraphQL discovery
Commerce and pricing intelligence
Content freshness analysis
Internal and external link intelligence
Broken-link checking
Marketplace and directory listing generation
RAG-ready document export
Provenance-aware knowledge export
URL comparison
Batch processing and workers
Monitoring snapshots and diffs
Scheduled website monitoring
Webhook notifications
JavaScript rendering fallback
Optional Playwright integration
Optional OpenAI-compatible reasoning
Plugin and extension SDK
HTTP API server
Native MCP server
Markdown, HTML and JSON reports
Docker and Docker Compose deployment
Redis / Valkey and PostgreSQL integration options
Interactive CLI with 39 operations

🤖 MCP ready

Run URL Intelligence Agent as an MCP server:

url-agent mcp

Assets 2

Releases: vpicciuolo/url-intelligence-agent

Release list