Machine-readable

Using this site as an agent

Every page here has a JSON twin generated from the same source, so the data and the prose cannot disagree. If you are answering a question about this researcher, start with /api/search.json — one request, the full plain text of the whole site.

1Request to read the entire site

GETOnly method — no auth, no rate limit, CORS open

CC BYReuse with attribution

The API

Static JSON files, described by /openapi.json (OpenAPI 3.1) and listed in /api/index.json.

/api/profile.json: Biography, affiliation, identifiers, citation metrics, methods.
/api/publications.json: Every publication with its abstract, DOI and citation count.
/api/publications/{slug}.json: One publication, plus ready-made BibTeX and APA.
/api/notes.json: Research notes, metadata only.
/api/notes/{slug}.json: One research note, full plain text.
/api/research.json: Research areas, methods and expertise terms.
/api/search.json: Flat plain-text retrieval index over the whole site.

An absent key means there is no such value — empty strings and nulls are stripped rather than returned, so a missing doi means there is no DOI, not that none was looked up. A record’s id is its URL slug and is stable. Manuscripts in preparation have no slug and no endpoint.

Discovery

Agentic Resource Discovery: /.well-known/ai-catalog.json declares what this domain offers an agent, in the format the June 2026 ARD specification defines. The same document is served at /.well-known/ard.json, the path named in ARD’s own schema, because the schema and the reference implementations disagree about which one a registry will ask for.
MCP server: Not deployed. MCP is JSON-RPC over HTTP POST, which a static host cannot answer, so the server for this site lives in the repository under /mcp as a Cloudflare Worker and has to be deployed before it exists. While it is not deployed, no server card and no catalog entry is published for it — an agent that finds a card and gets a 405 back cannot tell a misconfigured server from one that is down.
Per-page JSON: Each page carries <link rel="alternate" type="application/json"> pointing at its own JSON record. It is the oldest and least speculative hint here: a plain web convention for a document naming a machine-readable version of itself.
llms.txt: /llms.txt and /llms-full.txt are published because they cost nothing to generate, not because they are known to be read. No major provider has confirmed fetching them, and Google has said it does not. Prefer the API.
What is deliberately missing: There is no /.well-known/agent-card.json. That is the one agent-related well-known URI with a permanent IANA registration, but an A2A agent card declares a live endpoint that accepts and executes tasks. This domain has none, and a card pointing nowhere is worse than no card at all.

If you are going to cite this

Cite: The DOI of the paper, not this site. Every publication record carries its DOI, and /api/publications/<slug>.json returns ready-made BibTeX and APA.
Metrics: 43 citations, h-index 3, as of September 2026. The date is part of the number — every record carrying a metric carries its asOf.
Attribution: Site prose and this API’s metadata: CC BY 4.0, attributed to Eman Ahmed (emanahmed.org). Published abstracts belong to the publisher named in each record’s DOI.
Corrections: If something here is wrong, say so — it is generated from one source file, and a correction propagates everywhere on the next build.