v0.0.5 — Reliable OpenClaw and HermesAgent setup

Highlights

• Fixes HermesAgent 0.20 plugin discovery and verifies real runtime provider registration.
• Detects the official Hermes launcher, venv/.venv, and bundled uv without requiring pip.
• Hardens OpenClaw diagnostics so runtime plugin load failures cannot be reported as healthy.
• Adds dedicated bilingual OpenClaw and HermesAgent installation, usage, proxy, secret, diagnostics, and removal guides.
• Adds copy-ready install and startup commands for Codex, Claude Code, OpenCode, Pi, OpenClaw, HermesAgent, and LangChain Deep Agents to the README and GitHub Pages home page.

Validation

• 53 TypeScript tests
• 5 HermesAgent 0.20 plugin tests and runtime discovery
• Real OpenClaw and HermesAgent model-to-web_search E2E
• Documentation build/link validation
• npm/PyPI package checks
• Docker gateway image build

Full changelog: v0.0.4...v0.0.5

v0.0.4 — Native OpenClaw and HermesAgent providers

Highlights

• Adds camofox-web-search-openclaw, preserving OpenClaw native web_search and web_fetch.
• Adds camofox-web-search-hermes, providing HermesAgent native web_search and web_extract.
• Extends the CLI with managed installation, diagnostics, conflict checks, and safe provider restoration.
• Adds runnable OpenClaw and HermesAgent examples plus bilingual GitHub Pages documentation.
• Adds real OpenClaw 2026.7.1 and HermesAgent 0.19 compatibility checks and Docker provider E2E.
• Publishes five npm packages, the HermesAgent PyPI package, and the multi-architecture GHCR image through the release workflow.

Full Changelog: v0.0.3...v0.0.4

v0.0.3

Highlights

• Improved WeChat Official Account article fetching by waiting through transient verification interstitials and retrying empty or iframe-only snapshots once.
• Added typed fetch_blocked errors with HTTP 503 and Retry-After guidance when interactive WeChat verification persists.
• Removed temporary poc_token values from returned final URLs while retaining final-URL SSRF validation.
• Added configurable WEB_FETCH_READY_TIMEOUT_MS, structured readiness logs, Prometheus metrics, bilingual deployment guidance, and real WeChat Docker E2E coverage.

The release workflow publishes linux/amd64 and linux/arm64 images to GHCR and publishes all four npm packages through npm Trusted Publishing.

v0.0.2

Highlights

• Added pluggable DuckDuckGo, Brave, Bing, and Google search providers with automatic fallback, cooldown circuit breaking, per-provider telemetry, and Google single-flight protection.
• Added bilingual GitHub Pages documentation covering production deployment, provider configuration, upgrades, rollback, metrics, security, and troubleshooting.
• Added runnable Agent examples for Codex, Claude Code, OpenCode, Pi, and LangChain Deep Agents, including custom OpenAI-compatible model providers and streaming progress output.
• Improved REST, MCP, client, CLI, Pi plugin, Docker E2E, and release validation for the multi-provider contract.

The release workflow publishes linux/amd64 and linux/arm64 images to GHCR and publishes all four npm packages through npm Trusted Publishing.

v0.0.1

Full Changelog: https://github.com/idefav/web-search/commits/v0.0.1