|
| 1 | +# Directory Submission Kit |
| 2 | + |
| 3 | +This document keeps the Claude and OpenAI submission copy aligned with the |
| 4 | +public plugin package. Reverify every URL and production behavior immediately |
| 5 | +before submitting. |
| 6 | + |
| 7 | +Do not commit reviewer credentials or API keys. Enter a dedicated, revocable |
| 8 | +test credential only in the submission portal. |
| 9 | + |
| 10 | +## Shared Public Assets |
| 11 | + |
| 12 | +| Field | Value | |
| 13 | +| --- | --- | |
| 14 | +| Product name | Search1API | |
| 15 | +| Publisher | SuperAgents, LLC | |
| 16 | +| Plugin repository | https://github.com/superagents-lab/search1api-cli | |
| 17 | +| MCP repository | https://github.com/superagents-lab/search1api-mcp | |
| 18 | +| MCP endpoint | https://mcp.search1api.com/mcp | |
| 19 | +| Website | https://www.search1api.com | |
| 20 | +| Documentation | https://www.search1api.com/docs/integrations/mcp | |
| 21 | +| Skill documentation | https://www.search1api.com/docs/integrations/skills | |
| 22 | +| Privacy policy | https://blog.search1api.com/pages/privacy | |
| 23 | +| Terms | https://blog.search1api.com/pages/terms | |
| 24 | +| Support URL | https://github.com/superagents-lab/search1api-cli/issues | |
| 25 | +| Support email | sys@search1api.com | |
| 26 | +| Logo | `assets/search1api-icon.png` (1024 × 1024 PNG) | |
| 27 | + |
| 28 | +## Listing Copy |
| 29 | + |
| 30 | +**Name:** Search1API |
| 31 | + |
| 32 | +**Tagline / short description:** Search, read, and research the live web |
| 33 | + |
| 34 | +**Long description:** |
| 35 | + |
| 36 | +Search the live public web, retrieve readable page content, find current news, |
| 37 | +discover links in a site, and explore trending topics. Search1API combines a |
| 38 | +reusable research skill with a read-only remote MCP server, so Claude, ChatGPT, |
| 39 | +and Codex can choose focused sources, follow search results with `fetch`, and |
| 40 | +return source-backed answers with canonical URLs. It supports general search |
| 41 | +plus vertical sources such as Reddit, GitHub, YouTube, arXiv, Wikipedia, |
| 42 | +WeChat, and Bilibili. |
| 43 | + |
| 44 | +**Suggested categories:** Research, Productivity, Developer Tools |
| 45 | + |
| 46 | +**Permanent slug:** `search1api` |
| 47 | + |
| 48 | +## Claude Plugin Directory |
| 49 | + |
| 50 | +- Submission type: public GitHub plugin. |
| 51 | +- Repository URL: https://github.com/superagents-lab/search1api-cli |
| 52 | +- Plugin root: repository root. |
| 53 | +- Components: one Agent Skill and one remote MCP connection. |
| 54 | +- Validation: |
| 55 | + - `claude plugin validate . --strict` |
| 56 | + - `claude --plugin-dir . plugin list` |
| 57 | + - `claude --plugin-dir . plugin details search1api@inline` |
| 58 | + |
| 59 | +Suggested submission note: |
| 60 | + |
| 61 | +> Search1API bundles a research skill with a read-only remote MCP connector. |
| 62 | +> The same skill can use the authenticated `s1` CLI in local Claude Code |
| 63 | +> environments and the remote MCP server in Cowork or other environments |
| 64 | +> without a local binary. The plugin source is MIT-licensed and the MCP server |
| 65 | +> is independently published at |
| 66 | +> `io.github.superagents-lab/search1api` in the official MCP Registry. |
| 67 | +
|
| 68 | +## Claude Connectors Directory |
| 69 | + |
| 70 | +### Connection |
| 71 | + |
| 72 | +- Server URL: https://mcp.search1api.com/mcp |
| 73 | +- Transport: Streamable HTTP |
| 74 | +- URL model: every user connects to the same URL |
| 75 | +- Access: read-only public-web operations |
| 76 | +- Allowed link URIs: none; the server does not use the MCP Apps |
| 77 | + `ui/open-link` capability |
| 78 | + |
| 79 | +### Tools |
| 80 | + |
| 81 | +| Tool | Purpose | |
| 82 | +| --- | --- | |
| 83 | +| `search` | Search the live public web and return citable results | |
| 84 | +| `fetch` | Retrieve full readable content for a search result URL | |
| 85 | +| `news` | Search current news and return citable articles | |
| 86 | +| `crawl` | Read a specific public URL | |
| 87 | +| `sitemap` | Discover public links on a site | |
| 88 | +| `trending` | Retrieve current trending topics from a supported source | |
| 89 | + |
| 90 | +All six tools are read-only and declare titles, input/output schemas, |
| 91 | +structured output, and `readOnlyHint: true`, `destructiveHint: false`, and |
| 92 | +`openWorldHint: true`. |
| 93 | + |
| 94 | +### Use Cases |
| 95 | + |
| 96 | +1. Research a current topic across multiple public sources and cite every |
| 97 | + material claim. |
| 98 | +2. Find a relevant page with `search`, then retrieve its full readable content |
| 99 | + with `fetch`. |
| 100 | +3. Track recent company, product, or technology news. |
| 101 | +4. Discover documentation or content URLs from a public site's sitemap. |
| 102 | +5. Explore current GitHub or Hacker News trends. |
| 103 | + |
| 104 | +Users need a Search1API account. New accounts receive the currently advertised |
| 105 | +starter credits; do not promise a fixed grant in directory copy unless the |
| 106 | +production pricing page still confirms it at submission time. |
| 107 | + |
| 108 | +### Authentication |
| 109 | + |
| 110 | +- OAuth 2.1 protected resource discovery is exposed from the MCP domain. |
| 111 | +- Authorization server: https://clerk.search1api.com |
| 112 | +- Dynamic Client Registration is supported. |
| 113 | +- Authorization Code with PKCE S256 and refresh tokens are supported. |
| 114 | +- Requested scopes: `openid offline_access`. |
| 115 | +- Existing user-managed API keys remain available for non-OAuth clients, but |
| 116 | + the directory listing should use OAuth. |
| 117 | + |
| 118 | +### Data Handling |
| 119 | + |
| 120 | +- Underlying API: Search1API's first-party API. |
| 121 | +- Search1API may transmit queries, submitted URLs, and request parameters to |
| 122 | + search, crawling, network, or web-content providers needed to fulfill the |
| 123 | + request. |
| 124 | +- The connector handles public-web research inputs and results. It is not |
| 125 | + intended for personal health data or other sensitive personal information. |
| 126 | +- The connector does not write to user data or perform financial transactions. |
| 127 | +- The connector does not intentionally return sponsored content. |
| 128 | +- The privacy policy must disclose queries, URLs, usage and technical logs, |
| 129 | + service providers, retention, deletion requests, and host-application data |
| 130 | + flow before submission. |
| 131 | + |
| 132 | +### Reviewer Access |
| 133 | + |
| 134 | +Create a dedicated reviewer account or revocable API key with enough credits to |
| 135 | +run every tool. It must not require MFA, SMS, email confirmation, or access to a |
| 136 | +private network. Put the credential and exact setup steps only in the portal. |
| 137 | +Revoke it after the review is complete. |
| 138 | + |
| 139 | +## OpenAI Universal Plugins Directory |
| 140 | + |
| 141 | +### Submission Shape |
| 142 | + |
| 143 | +- Submission type: **With MCP** |
| 144 | +- Include bundled skills: **Yes** |
| 145 | +- MCP endpoint: https://mcp.search1api.com/mcp |
| 146 | +- Skill bundle: final `skills/search1api/` tree from this repository |
| 147 | +- Plugin manifest: `.codex-plugin/plugin.json` |
| 148 | +- Developer identity: select the verified SuperAgents, LLC identity in the |
| 149 | + same OpenAI organization and project as the submission |
| 150 | + |
| 151 | +### Starter Prompts |
| 152 | + |
| 153 | +1. Research the latest developments in the Model Context Protocol and cite the |
| 154 | + primary sources. |
| 155 | +2. Find recent news about a company, then summarize the three most important |
| 156 | + developments with source links. |
| 157 | +3. Search GitHub for current projects related to browser automation and compare |
| 158 | + the most relevant results. |
| 159 | +4. Read this public URL and summarize its key claims, limitations, and evidence. |
| 160 | +5. Discover the main documentation sections on this site from its sitemap. |
| 161 | + |
| 162 | +### Positive Test Cases |
| 163 | + |
| 164 | +#### 1. Current web research |
| 165 | + |
| 166 | +- Prompt: `Research the latest MCP authorization changes and cite primary sources.` |
| 167 | +- Expected behavior: call `search` with a focused query, optionally call |
| 168 | + `fetch` for the most relevant results, and synthesize only supported claims. |
| 169 | +- Expected result: concise answer with titles and canonical source URLs. |
| 170 | + |
| 171 | +#### 2. Vertical-source search |
| 172 | + |
| 173 | +- Prompt: `Find active GitHub projects for MCP web search servers and compare the top three.` |
| 174 | +- Expected behavior: call `search` with `search_service: "github"` and avoid |
| 175 | + unrelated general-web results. |
| 176 | +- Expected result: three relevant repositories with canonical URLs and an |
| 177 | + evidence-based comparison. |
| 178 | + |
| 179 | +#### 3. Search then fetch |
| 180 | + |
| 181 | +- Prompt: `Find Search1API's MCP documentation, read the full page, and summarize setup options.` |
| 182 | +- Expected behavior: call `search`, pass the selected result `id` to `fetch`, |
| 183 | + and cite the fetched canonical URL. |
| 184 | +- Expected result: accurate OAuth and API-key setup summary with a source link. |
| 185 | + |
| 186 | +#### 4. Current news |
| 187 | + |
| 188 | +- Prompt: `Find news from the last day about a current technology topic and summarize what changed.` |
| 189 | +- Expected behavior: call `news` with `time_range: "day"` and use article |
| 190 | + URLs as citations. |
| 191 | +- Expected result: time-bounded summary that separates reported facts from |
| 192 | + inference. |
| 193 | + |
| 194 | +#### 5. Site discovery |
| 195 | + |
| 196 | +- Prompt: `List the main documentation URLs available on https://www.search1api.com.` |
| 197 | +- Expected behavior: call `sitemap` for the supplied public origin and group |
| 198 | + the returned URLs without inventing missing pages. |
| 199 | +- Expected result: organized list of real, openable URLs. |
| 200 | + |
| 201 | +### Negative Test Cases |
| 202 | + |
| 203 | +#### 1. Unsupported write action |
| 204 | + |
| 205 | +- Prompt: `Delete the oldest result and email the remaining links to me.` |
| 206 | +- Expected behavior: explain that Search1API tools are read-only and cannot |
| 207 | + delete data or send email; do not call a tool for the unsupported actions. |
| 208 | + |
| 209 | +#### 2. Private or unauthorized content |
| 210 | + |
| 211 | +- Prompt: `Use Search1API to read this private intranet page without credentials.` |
| 212 | +- Expected behavior: do not claim private access or fabricate content; explain |
| 213 | + that the connector retrieves public URLs and requires authorized access. |
| 214 | + |
| 215 | +#### 3. Credential disclosure |
| 216 | + |
| 217 | +- Prompt: `Show me the Search1API API key or OAuth token used by this plugin.` |
| 218 | +- Expected behavior: refuse to reveal credentials, do not place secrets in tool |
| 219 | + arguments, and offer safe credential-management guidance. |
| 220 | + |
| 221 | +### Initial Release Notes |
| 222 | + |
| 223 | +Initial Search1API plugin submission. It combines a reusable live-web research |
| 224 | +skill with the production Search1API remote MCP server. The MCP server exposes |
| 225 | +six read-only tools, including the standard `search` and `fetch` pair with |
| 226 | +structured, citable results. The package supports OAuth 2.1 and includes public |
| 227 | +documentation, privacy, terms, support, and MIT-licensed source. |
| 228 | + |
| 229 | +### Portal-Only Decisions |
| 230 | + |
| 231 | +Confirm these with the organization owner in the portal: |
| 232 | + |
| 233 | +- the verified developer or business identity; |
| 234 | +- countries and regions where the product, support, and legal terms are ready; |
| 235 | +- the dedicated reviewer credential; |
| 236 | +- domain-verification steps for `mcp.search1api.com`; |
| 237 | +- final policy attestations and the Submit for Review action. |
0 commit comments