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

# Pagination and limits

> Current collection semantics, hard result caps, concurrency, and input bounds.

The v1 API does not expose a universal cursor, page, or offset contract. Do not send generic pagination parameters to every collection.

## Collection behavior

| Collection           | Current behavior                                                              |
| -------------------- | ----------------------------------------------------------------------------- |
| Credit usage         | <code>limit</code> defaults to 100 and has a maximum of 500                   |
| Page evaluations     | Returns the latest 20                                                         |
| GEO time series      | Uses the registered 7-day, 28-day, or all-history range                       |
| Dashboard insights   | Service returns up to 30; dashboard displays 12                               |
| Most workspace lists | Return the current authorized workspace collection without generic pagination |

The Billing UI requests up to 500 usage records and paginates 25 at a time in the browser. Those page numbers are not server cursors.

## Job concurrency

| Job                                                               | Limit                                           |
| ----------------------------------------------------------------- | ----------------------------------------------- |
| Measurement, research, recommendation generation, and SEO refresh | One shared running job slot per workspace       |
| Prompt regeneration                                               | One running prompt-generation job per workspace |
| Page evaluation                                                   | One running job per workspace                   |
| Research                                                          | One on-demand start per workspace per hour      |

The shared slot means a running SEO refresh can block a measurement start, for example. A prompt-regeneration or page-evaluation job uses its own kind-specific guard. An active-job conflict returns HTTP 409. The research start limit can return HTTP 429.

## Measurement caps

| Preset   |                     Active prompt cap |
| -------- | ------------------------------------: |
| Smoke    |                                     2 |
| Lite     |                                    50 |
| Standard | 150, with lower caps for some engines |
| Exa      |                                   150 |

The CLI-only Deep preset has a 150-prompt cap and three repeats.

## Workspace and prompt bounds

| Input                   | Limit                                         |
| ----------------------- | --------------------------------------------- |
| Workspace slug          | 2 to 20 lowercase letters, digits, or hyphens |
| Workspace display name  | 2 to 60 characters                            |
| Owned domains           | 1 to 8                                        |
| Regions                 | Up to 8                                       |
| Competitors at creation | 2 to 12                                       |
| Active prompt target    | 20 to 300                                     |
| Prompt text             | 3 to 2,000 characters                         |
| Prompt topic            | 1 to 120 characters                           |
| Prompt importance       | 0 through 1                                   |

## Other input bounds

| Input                       | Limit                                 |
| --------------------------- | ------------------------------------- |
| Thread title                | Up to 100 characters                  |
| API key name                | Up to 200 characters                  |
| Common resource identifiers | Up to 200 characters where registered |
| Stored credential value     | Up to 16,384 characters               |
| Design import               | Up to 50,000 characters               |
| Team display name           | Up to 120 characters                  |
| Competitor name             | Up to 60 characters                   |
| Competitor domains          | Up to 5                               |
| Competitor aliases          | Up to 8                               |

<Note>
  Route schemas remain authoritative. A generated operation can apply a narrower enum or format in addition to the limits summarized here.
</Note>

## Metered operation ceilings

These limits come from the current versioned pricing configuration. They cap authorization; they do not promise that every operation will use the full amount.

| Operation                | Current ceiling                                                        |
| ------------------------ | ---------------------------------------------------------------------- |
| Chat                     | 6 model steps, 2,048 output tokens and 120,000 micro-USD per step      |
| Prompt generation        | 100 provider calls, 8,192 output tokens and 250,000 micro-USD per call |
| Research                 | 80 provider calls with provider-specific token or result caps          |
| Pipeline technical audit | 21 URLs                                                                |

## Recommendation generation caps

One SEO generation operation creates at most 20 items toward a target of 50 open items. One GEO operation creates at most 10 toward a target of 20. Open means Proposed, Todo, or In review.

These are volume controls, not pagination.