Update to Pendar

{
  "name": "survey-a-research-field",
  "description": "Size and characterise a research area from exact publication counts — how many papers exist, whether output is growing or fading, which sub-areas and disciplines it divides into, which countries and institutions produce it, where it is published, and what to read first. Use when the user asks how large, how active, or how crowded a field is, wants an overview of a topic's literature, or is deciding whether to enter an area.",
  "included_files": [
    {
      "relative_path": "agents/openai.yaml",
      "size_in_bytes": 523
    },
    {
      "relative_path": "assets/pendar-logo.png",
      "size_in_bytes": 127623
    }
  ],
  "skill_md_contents": "---\nname: survey-a-research-field\ndescription: Size and characterise a research area from exact publication counts — how many papers exist, whether output is growing or fading, which sub-areas and disciplines it divides into, which countries and institutions produce it, where it is published, and what to read first. Use when the user asks how large, how active, or how crowded a field is, wants an overview of a topic's literature, or is deciding whether to enter an area.\nmetadata:\n  short-description: Measure a research field by exact publication counts\n---\n\n# Survey a research field\n\nDescribe a body of literature by counting it. `survey_literature` returns exact\ncounts over every paper matching the topic — not a sample — so the answer can\nstate how many papers exist, how that number moved year by year, and who\nproduced them.\n\nThe counts are the substance here. A list of relevant papers comes back too, but\nit is the smaller half of the answer, and leading with it wastes what the tool is\nfor.\n\n## Quick start\n\n1. Reduce the user's question to a **short noun phrase of two to six words**.\n   \"How much work has been done on solid-state electrolytes for lithium metal\n   batteries?\" becomes `solid state electrolytes lithium metal`.\n2. Call `survey_literature` once. Add `year_from` or `year_to` only if the user\n   asked about a period.\n3. Read `corpus.total_papers` together with `corpus.matched_on` before saying\n   anything about size — the second changes what the first means.\n4. Lead with size and trend, then composition, then the reading list.\n5. Repeat every caveat in `notes` that bears on what you claim.\n\n## Choosing the query\n\nThe search requires **every** word to appear, so each extra word shrinks the\ncorpus, and question words are stripped before searching.\n\n- Drop question framing, verbs, and articles: keep the subject.\n- Keep the phrase specific enough to mean one field and no more.\n- If `total_papers` is 0 or in the low single digits, the terms were too narrow.\n  Retry once with the two or three most central words, and tell the user you\n  broadened it and to what.\n- If the number looks implausibly large for the subject, the terms were too\n  generic. Retry with the distinguishing term restored.\n\nNever present a retry as a different question. State which phrase produced the\nnumbers you are reporting; `corpus.searched_for` shows what was actually\nsearched after stripping.\n\n## Reading the result\n\n- `corpus.total_papers` with `corpus.matched_on`:\n  - `title_and_abstract` — a conservative count. Papers that mention the topic\n    only in passing are excluded.\n  - `full_text` — too few papers had the topic in title or abstract, so the\n    corpus was widened to anything whose full text mentions it. Report the total\n    as an **upper bound**, because passing mentions are now included.\n- `corpus.trend` — `interpretation` is already written from the exact per-year\n  counts; use it rather than recomputing a trend from `papers_by_year`. Treat\n  `too_new` as insufficient history, not as a lack of growth. The current year is\n  partial, so never describe a field as declining on the strength of the latest\n  bar alone.\n- `corpus.topics` versus `corpus.disciplines` — topics are the **sub-areas the\n  work divides into**; disciplines are the **fields whose researchers produce\n  it**. Keep them apart. A topic list answers \"what is this field about\", and a\n  discipline list answers \"who works on it\".\n- `corpus.countries` and `corpus.institutions` — shares overlap and total well\n  above 100%, because a paper written across four institutions counts once in\n  each. Say so if you quote them.\n- `corpus.venues` — preprint servers and repositories rank here beside journals.\n  A repository near the top is a finding: the field circulates before peer\n  review.\n- `corpus.work_types` — a large preprint share means a fast-moving field where\n  the published record lags.\n- `corpus.retracted_papers` — mention any retractions if the user intends to rely\n  on the field's findings.\n- `papers` — the most relevant papers, with abstract snippets. Summarise what\n  they cover; do not present them as the field's most important work, since they\n  are ranked by relevance to the query rather than by influence.\n\n## Never\n\n- Read percentages from a small corpus as proportions of a research area. Exact\n  arithmetic over six papers still prints as \"33%\". When `notes` flags a small\n  corpus, describe the individual papers instead.\n- Compare two topics' totals without saying whether both were matched the same\n  way. A `full_text` total against a `title_and_abstract` total is not a\n  comparison.\n- Add papers, counts, or trends from memory. Everything reported must come from\n  the response.\n- Call the tool repeatedly to page through a field. One survey covers the whole\n  corpus; more calls only re-count it.\n\n## Reporting\n\n1. **Size** — the total, what it was matched on, and the years covered.\n2. **Trajectory** — the trend, with the recent versus prior three-year counts\n   behind it and the peak year.\n3. **Composition** — sub-areas, then the disciplines and countries producing the\n   work, then where it is published and how much is peer reviewed.\n4. **What to read** — the returned papers, grouped by what they address.\n5. **Caveats** — the corpus definition and any small-corpus or retraction note.\n\n## Related workflows\n\n- To measure one paper's influence rather than a field's size, use\n  `trace-citation-impact`.\n- To check a specific reference the user already has, use `verify-citations`.\n"
}