Getting Started
Connect ChatGPT, Codex and Scripts
Query public PlantQTLdb data from your own AI client without using the website AI allowance
Account and permissions
Register a PlantQTLdb account and verify your email. Active verified accounts can use read-only MCP data access automatically; administrator approval is not required.
MCP retrieves public gene annotations, molecular QTLs, stored trait associations and local literature evidence, with optional online publication discovery. Your own client handles interpretation. It cannot access your website conversations, uploaded files or account settings and does not use PlantQTLdb AI requests.
Connect ChatGPT
- Add the serverIn ChatGPT's custom MCP server settings, enter https://plantqtldb.cn/mcp and select OAuth authentication. Availability depends on your ChatGPT plan and workspace settings.
- Sign in and authorizeSign in with your PlantQTLdb account, check the requested read-only permission, and select Authorize read-only access. Return to ChatGPT to use the connection.
- Check the connectionAsk: Use PlantQTLdb to list supported species. Then provide a species and reference gene IDs for your query.
Connect Codex or scripts
For Codex CLI, run: codex mcp add plantqtldb --url https://plantqtldb.cn/mcp. Follow the sign-in and authorization prompts; if authorization does not start automatically, run: codex mcp login plantqtldb. OAuth connections do not require a personal token.
For scripts or clients that require a Bearer token, open Apps & API access in the PlantQTLdb Assistant user menu and create a named personal access token. Copy it when first shown; it expires in 90 days. Send it as an Authorization: Bearer header, not as a browser query parameter.
After server updates, an existing client may retain an older tool list. Refresh the connection's tools or reconnect PlantQTLdb, then open a new conversation to use newly added features.
Batch queries and access management
- Provide the species and a list of reference gene IDs. Start with gene summaries, then request the literature and association details you need; the client can process large lists in smaller batches.
- Request local literature for your gene list first. Save each batch and continue remaining genes or pages; a failed query is not zero evidence. Optional online discovery supplements saved local results, with source labels and gene/species mention checks. Merge duplicate papers by DOI, PMID or PMCID while preserving each gene-paper relationship. Online mentions are candidates, not confirmed experimental support.
- Standard limits are 60 data requests per minute and 5,000 per day per account across connected apps and tokens. Usage resets at midnight (UTC+8). These limits are separate from the AI allowance. Honor Retry-After on a 429 response.
- Open Apps & API access from the PlantQTLdb Assistant user menu to review app names, callback domains, read-only permissions and recent activity. Revoke an application or token to stop new requests immediately. Disabled or unverified accounts cannot use MCP.
- Read provenance, reference builds, pagination and truncation. Bounded association examples are not a complete export, and unavailable queries must not be interpreted as zero evidence.