---
name: landscape
description: "Use when the user asks for an overview of a research field, wants to understand a domain's key players, or asks 'what's happening in X' or 'give me a landscape of X'. Triggers on requests for field summaries, top authors in an area, or subdomain breakdowns."
---

# Field Landscape

An overview of a research field: size, growth, most-cited papers, most prolific authors, neighbouring fields.

Use the Valency Bond MCP tools available in the current host. Tool names may be qualified by an MCP server prefix; call the matching exposed tool. If the required Valency Bond tools are unavailable, say so and stop. If a tool call fails with an error, retry it once before treating it as a real failure.

## Input

- **arXiv category code** (`cs.LG`, `q-bio.NC`, `astro-ph.HE`) → category path.
- **Anything else** (free text, a PubMed topic, a bioRxiv label) → topic path. Only arXiv codes get the category path; other sources' labels go through free text.

A category call that returns an empty series with an `UNKNOWN_CATEGORY` warning means the code isn't recognised: show the suggested codes and ask which one was meant, never report a field with no activity. A `CATEGORY_TOO_BROAD` error lists the subcategories to use instead: show them and ask.

## Tool Chain — category path

1. **Most-cited papers:** `search_by_category` with `category`, `sort_by: "citations"`, `limit: 100`, `include_abstract: false`, `max_authors: 3`. The sort covers the whole category. Results include papers cross-listed into it; a paper's first-listed category is its primary one.
2. **Volume over time:** `get_publication_trends` with `category`, `granularity: "year"`, `format: "compact"`.
3. **Prolific authors:** `identify_prolific_authors` with `category`, `limit: 10`. Rows are keyed by name; use each row's identity fields:
   - `unambiguous`, `mostly_single`, or `ambiguous` with `top_candidate_coverage` ≥ 0.85 → one person.
   - other `ambiguous` → "may combine a few people with this name".
   - `likely_aggregation` / `severe_aggregation` → several people share this name; don't present the count as one person's.
   - `unresolved` → not enough signal to tell.
   Drop rows that are clearly not a person (a country, institution or collaboration name).
   Keep `top_candidate_id` for follow-ups. If the call times out, retry once, then skip the section and say author ranking wasn't available.
4. **Size:** `analyze_corpus_metrics` with `category`.
5. **Neighbouring fields (no call):** count the other categories listed on Step 1's papers.

## Tool Chain — topic path

1. **Most-cited papers:** `semantic_search_papers` with `query`, `sort_by: "citations"`, `limit: 100`, `include_abstract: false`, `max_authors: 3`. This takes the 100 papers closest in meaning and orders those by citations, so it's "the most-cited of the closest matches", not the field's all-time list. Phrase the query the way the field describes itself.
2. **Volume over time:** `get_keyword_trends` with `granularity: "year"`, `format: "compact"` and the short topic term (e.g. "base editing" — not the descriptive sentence from step 1). Keep any scope the request narrows to ("antibiotic resistance" plus the organism it names, not "antibiotic resistance" alone) so the volume matches the papers listed. When that scope has a common abbreviation, run both spellings ("COPD" and "chronic obstructive pulmonary disease"): use the spelled-out series for the table and totals unless the field mostly writes the abbreviation ("CRISPR", "HIV"), and give the other's total in one line. It matches the words, not the exact phrase, so years before the topic existed can hold unrelated hits.
3. No author ranking, corpus metrics or neighbouring fields on this path — they need a category. Leave those sections out without comment.

## Reading the numbers

- **Missing years:** years with no papers are left out of a trend series, not returned as 0. Fill them with 0 between the first and last year shown before building the table or comparing years.
- **Trends** count papers by `dates.first_available`, when each first became publicly available. Base the trajectory on complete years. Mark the current year as partial; if it already exceeds last year's full count, say so as a partial-year count rather than a trend. An earlier year at least double its neighbours' average, with no known field event: describe it in one line as a spike, without suggesting a cause, rather than calling it growth.
- **Topic-path totals:** sum the keyword-trend counts from the year the topic took off (when counts start rising steadily), labelled "papers mentioning the term since <year>"; mention the earlier stray matches in one line.
- **Duplicates:** the same work can appear as an arXiv and a PubMed record. Treat rows as one paper when the DOI matches, or the title matches (ignoring case, punctuation and hyphens) **and** the first author matches. Keep the higher citation count.

## Displaying papers

Link each title to its `url`; keep the ID in parentheses. Year: the year in `journal_ref` when present, else the year of `dates.first_available`: of its `value` at `day`, `month` or `year` precision, or at `range` precision (`start` and `end`, no `value`) the year `start` and `end` share, none if they differ (where a row also carries `first_submitted`, it is the same value); if neither gives a year, leave the year off. Category: arXiv codes only — PubMed `categories` are free keywords. Authors: all if three or fewer, else first three + "et al."; a collaboration name ("The LIGO Scientific Collaboration") counts as an author.

## Output Format

### Field Summary
Total papers (category: `analyze_corpus_metrics`; topic: trend sum), years covered, growth in one or two sentences.

### Publication Trends
Year-by-year table, current year marked partial, spikes marked. For long series, group the sparse early years into one row.

### Most-Cited Papers
Top 10 (after merging duplicates): title (linked, ID), authors, year, citations. Category path: if cross-listed papers dominate the top 10, list the top papers whose primary category *is* the input first, then a short "also highly cited, cross-listed from other fields" line. Topic path: label the list "most-cited among the 100 closest matches", and, when the request narrows the topic to a specific disease, organism or setting ("… in children", "… in tuberculosis"), leave out papers whose title is clearly about a different one (keep general or method papers), moving down the list to fill 10 (show fewer if fewer remain) and saying in one line how many were left out.

### Top Authors (category path only)
Up to 10: name, paper count, with the identity note from Step 3 where one applies.

### Neighbouring Fields (category path only)
Most frequent co-listed categories with counts, labelled "co-listed on the field's 100 most-cited papers".

### Observations
2–3 points drawn only from the data above.

### Suggested Follow-ups
- Ask for a profile of `<author>` — category path only, for a listed Top Authors row (pass that row's `top_candidate_id` as the author_id)
- Ask for trends for `<category or topic>` — finer trend analysis
- Ask for papers similar to `<paper_id>` — around any listed paper