Context Window Optimiser

Rank document chunks by relevance and choose the ones that fit your token budget, with a score for each.

Pay by wallet (x402) AI-generated context-window-optimize

Example

Input

{
  "query": "SELECT * FROM users WHERE active = true LIMIT 10",
  "documents": [
    "test_item"
  ]
}

Answer

{
  "query": "SELECT * FROM users WHERE active = true LIMIT 10",
  "excluded": [
    {
      "reason": "low relevance"
    }
  ],
  "selected": [],
  "max_tokens": 4000,
  "selected_count": 0,
  "total_documents": 1,
  "tokens_remaining": 4000,
  "total_tokens_used": 0
}

Inputs

Field	Type	Required	What it is
`query`	string	Yes	—
`documents`	array	Yes	Array of {id, text}
`max_tokens`	integer	No	—

Output fields

How reliably each field appears in a successful answer.

Field	Type	Appears
`excluded`	array	Always (present in every successful answer)
`selected`	array	Always (present in every successful answer)
`tokens_remaining`	integer	Always (present in every successful answer)
`total_tokens_used`	integer	Always (present in every successful answer)

How to call it

The same tool through each route. MCP clients reach it through Strale’s server at https://api.strale.io/mcp; the call below is what the agent sends.

# Needs an API key. max_price_cents caps what this call may cost.
curl -X POST https://api.strale.io/v1/do \
  -H "Authorization: Bearer $STRALE_API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{
    "capability_slug": "context-window-optimize",
    "inputs": {
      "query": "SELECT * FROM users WHERE active = true LIMIT 10",
      "documents": [
        "test_item"
      ]
    },
    "max_price_cents": 3
  }'

{
  "tool": "strale_execute",
  "arguments": {
    "slug": "context-window-optimize",
    "inputs": {
      "query": "SELECT * FROM users WHERE active = true LIMIT 10",
      "documents": [
        "test_item"
      ]
    },
    "max_price_cents": 3
  }
}

import { Strale } from "straleio";

const strale = new Strale({ apiKey: process.env.STRALE_API_KEY! });
const result = await strale.do({
  capability_slug: "context-window-optimize",
  inputs: {
    "query": "SELECT * FROM users WHERE active = true LIMIT 10",
    "documents": [
      "test_item"
    ]
  },
  max_price_cents: 3,
});

import os
from straleio import Strale, DoRequest

strale = Strale(api_key=os.environ["STRALE_API_KEY"])
result = strale.do(DoRequest(
    capability_slug="context-window-optimize",
    inputs={
        "query": "SELECT * FROM users WHERE active = true LIMIT 10",
        "documents": [
            "test_item",
        ],
    },
    max_price_cents=3,
))

# No account. The first request answers 402 with the price in USDC;
# an x402 client pays it from your wallet and repeats the request.
curl -i -X POST https://api.strale.io/x402/v2/context-window-optimize \
  -H 'Content-Type: application/json' \
  -d '{
    "query": "SELECT * FROM users WHERE active = true LIMIT 10",
    "documents": [
      "test_item"
    ]
  }'

Limitations

Token savings depend on the target model
Optimisation suggestions use general practices; actual token savings depend on the target model’s tokenizer.
What to do instead: Measure actual token counts with the target model's tokenizer before and after applying suggestions