v0.18.13

Changed

@agentforge/tools - Concentrated HTML Document Extraction

• Centralized HTML parsing, element extraction, link/image traversal, URL resolution, and metadata extraction in one private document module shared by Web Scraper, HTML Parser, Extract Links, and Extract Images. Each invocation parses its HTML once.
• Preserved public factories, exports, schemas, defaults, Tool metadata, result shapes, and published TypeScript inference, including existing permissive result types.
• Preserved ordering, duplicates, empty and missing attributes, text and HTML selection behavior, URL-base selection and resolution fallback, metadata precedence, HTTP request handling, and destination checks.
• Added public Tool characterization tests and generated declaration/package-surface compatibility coverage for spec #314.

Validation

• pnpm release:validate passed with 251 test files passed, 2 skipped, 3,202 tests passed, and 23 skipped.
• Skips are intentional opt-in coverage for the Neo4j integration service (13 tests), PostgreSQL connection tests (4 tests), and web-search performance tests (6 tests); enable them with a running Neo4j service and RUN_INTEGRATION_TESTS=true, POSTGRES_CONNECTION_STRING, or RUN_WEB_PERFORMANCE_TESTS=true plus network access, respectively.

Published

• All packages published to npm registry at version 0.18.13:
  • @agentforge/core@0.18.13
  • @agentforge/skills@0.18.13
  • @agentforge/patterns@0.18.13
  • @agentforge/tools@0.18.13
  • @agentforge/testing@0.18.13
  • @agentforge/cli@0.18.13