SEARCH & RESEARCH · FOR AI AGENTS
OpenAlex
Find scholarly works, resolve citations and inspect research connections through API or MCP.
OpenAlex gives research agents access to a maintained scholarly graph covering works, authors, institutions and citations. Its MCP connector returns reproducible queries and can resolve references; the REST API supports more detailed retrieval and pagination.
Your integration checklist
- Capabilities
- What you need from your owner
- Connect MCP and approve OpenAlex sign-in, or obtain a free API key. An agent can request a key for its owner, but the owner must click the emailed activation link before that key works.
- Authentication
- MCP at https://mcp.openalex.org/mcp uses OAuth over Streamable HTTP. REST supports limited keyless access and verified API keys. Keep keys in the agent’s secret store.
- Your first task
- Resolve a small bibliography into work IDs and DOIs, retain match confidence, then inspect any ambiguous references before citing them.
- Billing & payment
- The free account includes a daily usage budget without a payment method. Paid prepaid usage and annual plans are optional. This listing’s no-payment route is the free allowance; usage resets at midnight UTC.
- Agent payment support
- No payment method is required for the free daily OpenAlex allowance. Optional prepaid usage and annual plans need a separate billing decision; they are not included in this no-payment access route.
- Operational limits
- The MCP connector provides metadata and links, not PDF retrieval or bulk export. Citation counts are not evidence-quality ratings. Coverage and author matching can be incomplete; profile-editing tools require separate authorization.
ORIGINAL AI AGENT STORE INTEGRATION PLAYBOOK
Keep a reproducible search trail before summarizing research
Editorial recommendation · Reviewed 2026-11-03
When to choose this service
Choose OpenAlex for scholarly discovery, reference checking and citation relationships. Read the underlying papers when the task needs methods, results or full-text evidence.
A maintained graph of publications, identifiers and citation links is an external research resource. An agent can analyze the records but cannot reliably invent missing references or keep the corpus current through prompting.
Suggested workflow
- Define the topic, date range and inclusion criteria, then preview a query and inspect a sample. Save the exact query so another agent can reproduce the search.
- Resolve candidate works using identifiers and inspect uncertain matches. Keep publication type, retraction status and source links alongside the records.
- Read the relevant papers before drawing substantive conclusions. Use API pagination or the documented export route for larger datasets, while staying within the chosen budget.
Acceptance checks before real work
- A matching title alone is insufficient when authors, year or DOI disagree.
- Highly cited work can still be irrelevant, outdated or retracted.
- A metadata record or abstract is not equivalent to having read the full paper.
If the workflow fails
On budget exhaustion, preserve progress and wait for the reset or obtain an approved upgrade. For an ambiguous citation, return the competing matches and missing evidence instead of asserting a fabricated DOI.