> ## Documentation Index
> Fetch the complete documentation index at: https://docs.iqrush.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Retrieve complete results

> Different tools limit results in different ways. Use the mechanism the tool actually provides, and tell report readers when you have returned only part of the data.

Different tools limit results in different ways. Use the mechanism the tool actually provides, and tell report readers when you have returned only part of the data.

| Tool or family | Mechanism | Completion check |
| - | - | - |
| Measurement leaderboards | `subject` selection and `top_n` | Inspect `subject_summary`; use `kind: all` when the complete set is needed. |
| `list_mention_statements` | Opaque `cursor`; `limit` 1–100 | Pass `data.cursor` into the next call until absent or null. |
| `get_placement_targets` | `offset`; `limit` 1–200 | Continue with `next_offset` until null. |
| `list_queries` | `limit` 1–200 | Check `data.n_total_run` and `subject_summary`; there is no cursor argument. |
| Time-series tools | `top_n` 1–200 | This selects series, not pages. Movers also limits returned rises and falls. |
| Google report history | `limit` 1–24 | This selects historical snapshots, not individual events. |

## Page through statements

Keep the project, run, entity selection, filters, and text settings the same on each call. After a successful page, copy `data.cursor` unchanged into the next request’s `cursor` argument. Stop if the server returns no next cursor. If a cursor repeats, stop and surface the issue instead of looping indefinitely.

Multiple `entities` selects statements mentioning any of those entities. It does not require every entity to appear together. Use `get_comention_network` for co-mention counts.

## Count the right thing

For queries, `n_total_run` is the total that ran, `n_available` is the matching selection, and `n_returned` is the number returned. Do not report a 50-row sample as “50 queries ran.” If a query selection is larger than the tool's limit, it is truncated. Do not invent an offset parameter to fetch the rest.

A statement is a returned unit of answer text, such as a paragraph, list item, table row, or heading. It can contain more than one sentence. Statement counts, raw term occurrences, and whole-response counts therefore need different labels.

Request full response text only when you need it. `max_text_chars` is limited to 50–2000, so even an included response can be shortened. Preserve the source reference when quoting evidence.

Next: [Errors and retries](/developers/guides/13-errors).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.