AnySearch Skill
Unified real-time search engine skill for AI agents, supporting web search, vertical search, batch queries, and full-page extraction.
- Source repo
- anysearch-ai/anysearch-skill
- Stars
- ★ 6.6k
- Last updated
- 1mo ago
- License
- Apache-2.0
- Primary language
- Python
- FA score
- 48/100 · Major gaps
At a glance
- How it runs
- Works with
- Universal · cross-platformCodex · Claude Code · OpenAI API · Claude API
- Cost
- Free tier plus a paid hosted plan
- Setup effort
- Low · running in minutes
- You'll need
- Typical use
- A developer using Claude Code needs to quickly find the latest documentation for an API; they use the
searchcommand to get relevant results. - Not a fit if
- Teams that need fully offline or air-gapped operation
- Users unwilling to depend on AnySearch's proprietary API and account
- Anyone needing PDF, Office, image, or media extraction
- Source review
- 48/100 · Major gaps 1 safety controls not found
What does this agent do, and when should you use it?
AnySearch Skill is a unified real-time search engine skill designed for AI agents, integrable into platforms like Claude Code, OpenCode, Cursor, and Windsurf. It provides general web search, vertical domain search (based on a built-in domain list), parallel batch search, and full-page content extraction. The skill is implemented as a multi-language CLI (Python, Node.js, PowerShell, Bash) with automatic runtime detection and configuration. An API key is optional but recommended; anonymous access has lower rate limits. After installation, agents use the CLI to perform searches and extractions, outputting JSON or Markdown.
The skill executes search and extraction operations by calling the AnySearch backend API. Agents use the CLI (e.g., anysearch_cli.py) for the following: search performs general web search with query and result limits; batch_search processes multiple queries in parallel; extract retrieves full-page content from a URL as Markdown. API keys can be obtained via a single API call using an email address; key priority is CLI flag, .env file, environment variable, then anonymous. The CLI supports Python, Node.js, PowerShell, and Bash, with recommended runtime auto-detected at install.
- A developer using Claude Code needs to quickly find the latest documentation for an API; they use the
searchcommand to get relevant results. - A researcher wants to gather information on multiple topics efficiently; they use
batch_searchto run queries in parallel. - A content writer needs to extract the body text of a webpage for reference; they use
extractto get Markdown output. - An agent developer wants to add search capability to a custom platform; they call the CLI or API directly, minimizing dependencies.
How do you install or deploy this agent?
To install: download the release archive via curl -L -o anysearch-skill.zip https://github.com/anysearch-ai/anysearch-skill/archive/refs/tags/v3.0.1.zip, unzip, and move to your agent's skill directory, e.g., mv anysearch-skill-3.0.1 ~/.claude/skills/anysearch. Configure the API key by copying .env.example to .env and setting ANYSEARCH_API_KEY, or obtain one via registration. After installation, run python <skill_dir>/scripts/anysearch_cli.py doc to detect the runtime and create runtime.conf.
How do you use this agent?
To use, invoke the CLI using the stored runtime command. For example: python <skill_dir>/scripts/anysearch_cli.py search "query" --max_results 5 for a basic search; python <skill_dir>/scripts/anysearch_cli.py batch_search --queries '[{"query":"q1","max_results":5},{"query":"q2","max_results":5}]' for batch processing; python <skill_dir>/scripts/anysearch_cli.py extract "https://example.com/page" to extract a page.
What are this agent's strengths and limitations?
- Multi-runtime support (Python/Node.js/PowerShell/Bash) with automatic detection of the recommended runtime for compatibility.
- Parallel batch search and full-page extraction enhance efficiency and comprehensiveness.
- Automated API key registration via email, without verification codes.
- Bilingual documentation (English and Chinese) and multi-language CLI implementations.
- Dependency on AnySearch proprietary API introduces vendor lock-in; anonymous usage has rate limits.
- Requires network connectivity; offline scenarios are not supported.
- Installation involves manual runtime configuration and API key setup, which may be a barrier for non-technical users.
How does this agent compare with similar options?
Key facts side by side with the most closely related agents.
| Agent | Source review | Form / cost | Stars | Updated | Language | Full support on |
|---|---|---|---|---|---|---|
| AnySearch Skill This agent | 48 · Major gaps | Agent plugin / skillFreemium | ★ 6.6k | 1mo ago | Python | Codex · Claude Code · OpenAI API · Claude API |
| wigolo | 54 · Major gaps | CLIFree + model costs | ★ 5.5k | 22d ago | TypeScript | Codex · Claude Code · OpenAI API · Claude API |
| Tavily Python SDK | 46 · Major gaps | Library / SDKFreemium | ★ 1.4k | 12d ago | Python | — |
| Lagent Lightweight Agent Framework | 43 · Major gaps | Library / SDKFree + model costs | ★ 2.3k | 6mo ago | Python | OpenAI API |
How does FollowAgents rate this agent?
Why each dimension lost points
Evidence shows: CLI scripts accept API keys via command-line arguments or environment variables, no hardcoded credentials found; registration flow requires a real email and clearly states a password is emailed, but lacks user confirmation step; documentation explains API key priority and storage, but does not detail data flow through third-party services; no special handling of sensitive data; dependency only requests>=2.20, no known vulnerabilities, but no dependency locking or audit; external effects include calling external API and sending registration requests, documented but without user confirmation; no rollback mechanism; source attribution only via README and LICENSE copyright notice, no contributor information. Deductions: missing user confirmation, insufficient data flow transparency, insufficient sensitive data handling, insufficient dependency security, external effects without user confirmation, no rollback, incomplete source attribution.
Evidence shows: README and SKILL.md descriptions are consistent, CLI scripts are generated by generate.py and kept in sync, CI checks script consistency; dependency requests>=2.20 is available from PyPI but not version-locked; error handling is well documented, including registration error branches and CLI failure prompts. Deductions: dependency not version-locked, may affect reproducibility; no specific runtime error examples.
Evidence shows: clearly targets AI agents and humans, supports multiple scenarios (search, batch search, extract); capability boundaries defined via subcommands and parameters; trigger conditions clear via SKILL.md and CLI help; environment adaptation supports Python, Node.js, PowerShell, Bash with runtime detection. Deductions: no detailed scenario examples or edge cases.
Evidence shows: clear file structure, README provides installation steps, naming stable (anysearch), example commands provided, but no FAQ; known limitations only mention anonymous rate limits, not comprehensive; license is Apache-2.0, version tags exist but no changelog; maintenance responsibility not explicit. Deductions: missing FAQ, incomplete known limitations, no changelog, unclear maintenance responsibility.
Evidence shows: output is JSON or Markdown, easy for agents to use; provides unique features like search, batch search, extract, high marginal value; cost-benefit: anonymous access rate-limited, API key free but quota not specified. Deductions: incomplete cost-benefit info, no performance or quota details.
Evidence shows: README claims consistent with code structure, but no external verification; no independent sources; documentation distinguishes facts and inferences but not explicitly. Deductions: lack of external verification and independent sources, fact-inference separation not explicit.
- Not found in source: rollback or recovery pathBack up first, or work on a git branch or snapshot, so its changes can be undone.
- Registration flow requires a real email and sends password to it, but lacks user confirmation step, raising privacy concerns.
- API key is written in plaintext to .env file; documentation recommends environment variables but no encryption at rest.
- Dependencies are not version-locked, affecting reproducibility and security.
- No rollback mechanism; misconfiguration may be hard to recover.
- Publisher identity unverified; source attribution incomplete.
Recent releases
Latest entries from the project's GitHub Releases. Notes are quoted from upstream as written, not rewritten or tested by FollowAgents.
-
Release notes excerpt
v3.1.1 — CLI Checksums & Automatic Client Version Sync 🚀 Update Your Agent Tell your agent to update the AnySearch skill: https://github.com/anysearch-ai/anysearch-skill Upgrade Option 1 - Tell your agent: ```txt Update the AnySearch skill: https://github.com/anysearch-ai/anysearch-skill ``` Option 2 - Run manually: ```bash git -C <your_skill_dir> pull ``` What's New in v3.1.1 🔐 CLI Integrity Verification - Added SHA256SUMS.txt with checksums for the Python, Node.js, PowerShell, and Bash CLI scripts …View on GitHub ↗ -
v3.1.0 v3.1.0 — Direct HTTP CLI, REST-native Search & Batch Improvements
Release notes excerpt
🚀 Update Your Agent Tell your agent to update the AnySearch skill: https://github.com/anysearch-ai/anysearch-skill --- Upgrade Option 1 - Tell your agent: ```txt Update the AnySearch skill: https://github.com/anysearch-ai/anysearch-skill ``` Option 2 - Run manually: ```bash git -C <your_skill_dir> pull ``` --- What's New in v3.1.0 ⚡ Direct HTTP CLI - Python, Node.js, PowerShell, and Bash CLIs now call /v1/search, /v1/sub-domains, and /v1/extract directly - Removed the MCP/JSON-RPC wrapper and improved API error reporting …View on GitHub ↗ -
v3.0.1 v3.0.1 — One-Step Registration, Flattened sub_domain_params & CLI Hardening
Release notes excerpt
🚀 Update Your Agent Tell your agent to update the AnySearch skill: https://github.com/anysearch-ai/anysearch-skill --- Upgrade Option 1 — Tell your agent: ```txt Update the AnySearch skill: https://github.com/anysearch-ai/anysearch-skill ``` Option 2 — Run manually: ```bash git -C <your_skill_dir> pull ``` --- What's New in v3.0.1 📧 One-Step Email Registration - Register an API key with a single call — no verification-code round-trip - All 4 CLI runtimes now send an X-Anysearch-Client: skill/3.0.1 header, version aligned with SKILL.md 🔧 Flattened sub_domain_params …View on GitHub ↗
FAQ
Is an API key required?
How do I get an API key?
What search types are supported?
How do I verify the skill works after installation?
doc command to probe runtimes, then perform a test search; a successful JSON response indicates a working connection.