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

> ## Agent Instructions
> Assistants connect to Parse over MCP at [https://agent.parse.gl/mcp](https://agent.parse.gl/mcp) (Streamable HTTP). With no sign-in it answers from Parse's public view (the numbers parse.gl's public pages show, up to 5 rows a list), starting with `check_brand`. Signed in with Parse, or sent a workspace key (`parse_mcp_…`), it works in the person's workspace, starting with `get_brief`. Scripts call the same tools over REST at `https://agent.parse.gl/v1/tools/<tool>`.
>
> ```bash
> npx parse-gl init
> ```
>
> To install just the skills: `npx skills add Parsegl/skills`.
> Every tool is also a POST to https://agent.parse.gl/v1/tools/<tool>; its OpenAPI description is https://agent.parse.gl/openapi.json.

# Read anything by id

> read_item: a workspace tool on Parse for agents (agent.parse.gl). Arguments, price, and a call over REST and MCP.

Open any Parse id returned by another call (one id kind per call): ans\_ (up to 10 AI answer texts with brands marked ⟦…⟧, judgments, citations, comparisons), pg\_ or URL (up to 20 with text, 100 without: stored copy, pageKind, brands listed, pitch check, robots verdicts; pass fields: \["contact"] for contacts at \$0.05 per newly unlocked route, free on misses and 90d repeats), exp\_ (export file page by cursor), or job\_/cf\_ (up to 25 async job statuses, costs, and results; wait: true holds until done). Excerpts free; full texts free for own content and 25 public texts/day (\$0.005 each after).

## Who may call it, and what it costs

| Name | Access | Effect |
| - | - | - |
| `read_item` | A workspace key, or a sign-in with read access. | Reads only |

**Price.** Free. A full text (`format: "full"`) of an answer or a page needs credits: the workspace's own answers and pages free, and public ones 25 a day free, then \$0.005 each, shown in `meta.costCredits`. A public answer older than 90 days costs \$0.001 the first time the workspace counts or lists it, and is the workspace's after, shown in `meta.costCredits`. An export's rows were counted when the file was written. Every reply says what the call cost in `meta.costCredits`. The prices are on [Spending, tracking and stopping](/docs/credits.md).

## Arguments

**ids** `list of string` *required*

ids of one kind: ans\_ (up to 10), pg\_ or URLs (20 with text, 100 when fields leaves text out), one exp\_, or job\_/cf\_ (up to 25) up to 100

**fields** `list of string`

the parts to return. Answers: the parts to return, besides each answer's id, date, engine and question; default all five. Without text, ten answers fit one compact reply. Pages: the parts of each page to return (ids always stay); a part left out is not loaded. Default omits "contact" (\$0). Pass fields: \["contact"] to unlock actionable contact routes (\$0.05 each, free on misses, bare author names, and 90d repeats) up to 12

**format** `one of compact, full`

full: whole text for up to 5 answers or 3 pages, free with credits within the day's allowance

**brands** `list of string`

answers and pages: brd\_ ids to mark, and on pages to answer for, besides the brands each names up to 20

**explain** `boolean`

false: leave out the one-sentence note (data.note)

**cursor** `string`

an export, or a job with call: page.cursor of the last page

**limit** `integer`

an export: rows in the page min 1, max 500

**wait** `integer`

jobs: seconds to wait for the jobs to finish, at most 50: the call returns as soon as none is queued or running or a first run's first look arrives, else when the time is up. Saves polling min 1, max 50

**call** `boolean`

jobs: with one job\_ id of an ask: the whole ask\_ai\_engines call that wrote it, its totals and its jobs 25 a page (cursor for the next); wait is not used

Every argument the tool takes, as `GET /v1/tools` serves it, every shared object written out; `https://agent.parse.gl/openapi.json` serves the same schema with the filter and the window named once under `components` and pointed to by `$ref`. MCP `tools/list` leaves out the arguments few calls send, and says the filter and the window once, under `search_answers`; `resolve_ids {"kinds": ["argument"]}` lists what it left out, and a call with any of them runs:

```json theme={null}
{
  "type": "object",
  "properties": {
    "ids": {
      "minItems": 1,
      "maxItems": 100,
      "type": "array",
      "items": {
        "type": "string",
        "minLength": 3,
        "maxLength": 2000
      },
      "description": "ids of one kind: ans_ (up to 10), pg_ or URLs (20 with text, 100 when fields leaves text out), one exp_, or job_/cf_ (up to 25)"
    },
    "fields": {
      "description": "the parts to return. Answers: the parts to return, besides each answer's id, date, engine and question; default all five. Without text, ten answers fit one compact reply. Pages: the parts of each page to return (ids always stay); a part left out is not loaded. Default omits \"contact\" ($0). Pass fields: [\"contact\"] to unlock actionable contact routes ($0.05 each, free on misses, bare author names, and 90d repeats)",
      "minItems": 1,
      "maxItems": 12,
      "type": "array",
      "items": {
        "type": "string",
        "maxLength": 40
      }
    },
    "format": {
      "description": "full: whole text for up to 5 answers or 3 pages, free with credits within the day's allowance",
      "type": "string",
      "enum": [
        "compact",
        "full"
      ]
    },
    "brands": {
      "description": "answers and pages: brd_ ids to mark, and on pages to answer for, besides the brands each names",
      "maxItems": 20,
      "type": "array",
      "items": {
        "type": "string",
        "minLength": 3,
        "maxLength": 220
      }
    },
    "explain": {
      "description": "false: leave out the one-sentence note (data.note)",
      "type": "boolean"
    },
    "cursor": {
      "description": "an export, or a job with call: page.cursor of the last page",
      "type": "string",
      "maxLength": 40
    },
    "limit": {
      "description": "an export: rows in the page",
      "type": "integer",
      "minimum": 1,
      "maximum": 500
    },
    "wait": {
      "description": "jobs: seconds to wait for the jobs to finish, at most 50: the call returns as soon as none is queued or running or a first run's first look arrives, else when the time is up. Saves polling",
      "type": "integer",
      "minimum": 1,
      "maximum": 50
    },
    "call": {
      "description": "jobs: with one job_ id of an ask: the whole ask_ai_engines call that wrote it, its totals and its jobs 25 a page (cursor for the next); wait is not used",
      "type": "boolean"
    }
  },
  "required": [
    "ids"
  ],
  "additionalProperties": false
}
```

## Example call

Read the text of answers and the brands each names, from ids a search returned. Over REST:

```bash theme={null}
export PARSE_API_KEY="<your-key>"

curl -s -X POST https://agent.parse.gl/v1/tools/read_item \
  -H 'content-type: application/json' \
  -H "Authorization: Bearer $PARSE_API_KEY" \
  -d '{"ids":["ans_…"],"fields":["text","brands"]}'
```

Over MCP, the assistant calls the tool by name with the same arguments:

```json theme={null}
{
  "method": "tools/call",
  "params": {
    "name": "read_item",
    "arguments": {
      "ids": [
        "ans_…"
      ],
      "fields": [
        "text",
        "brands"
      ]
    }
  }
}
```

More calls:

* Read the whole text: free with credits, within the day's full texts: `{"ids":["ans_…"],"format":"full"}`
* Who to write to about a page, for up to 100 pages: `{"ids":["pg_…"],"fields":["contact"]}`
* Read a page of the file an export wrote, through the service, with no fetch tool: `{"ids":["exp_…"],"limit":200}`
* The status and results of an ask or a baseline, holding the call up to 30 seconds until it finishes: `{"ids":["job_…"],"wait":30}`

The reply has `data`, `context`, `next`, `page` and `meta`; follow `next` for the calls that come after. A refused call returns `{ code, message, fix }`: see the [error table](/docs/mcp-reference.md#errors).


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