For developers

Your AI isthe frontend.

Stashlet is the backend. Add things by talking or typing to any AI, and ask for what you need when you need it. When you want your own pages, build them in any language on the same data.

Works from any language. The docs are written for your AI.

ChatGPTClaudeCATEGORYGroceriesDATE17NovPLACETescoAMOUNT£42AMOUNT£42$ curl /v1/query200 OK
Frontend: any AIBackend: your stashYour page, if you want one
Works with
  • Claude
  • ChatGPT
  • Claude Code
  • Codex
  • Cursor
  • GitHub Copilot
  • Windsurf
  • + any MCP client
One example, start to finish

A training plan with no app.

A progressive overload programme means lifting a little more each week. Nothing below is prebuilt. The AI sets it up from one sentence.

  1. ClaudeMonday
    You

    Put me on a bench press programme. Add 2.5 kg every week.

    • Created the workouts table
    • Set up a rule: +2.5 kg a week
    1

    Say what you want

    Claude creates a workouts table and sets up a rule for the weekly increase.

  2. ChatGPTMonday, 7pm
    You

    Bench, 3 sets of 5 at 60 kg.

    • Saved to workouts
    2

    Log it by voice

    Use whichever AI is open. Each detail lands in its own field.

  3. 60 kg + 2.5 kg = 62.5 kg

    Saved for next Monday

    3

    The rule plans next week

    When the row is saved, a simple formula works out the next session and saves it. It runs on our side with no AI. Your AI can also do the sum itself.

  4. Claude
    You

    What’s my workout today?

    Bench press, 3 × 5 at 62.5 kg.

    4

    Ask on the day

    No app, no website, nothing prebuilt. Any AI you connected can answer.

You don’t need to build a screen for every tool. When you want your numbers, ask your AI.

What you can build

Three things to build this week.

Each one is a few tables in one stash. Your AI does the typing.

  • Your own tools instead of subscriptions

    Workouts, a budget, a reading list. No forms to build: your AI works out which table and which fields. A new tool is just a new table in the same stash, with no new project, database or sign-in.

    You say

    “Just did legs, three sets of squats at 60 kg.”

    Then
    • A row in workouts: squats, 3 sets, 60 kg, today.
    • Your own progress page shows it, or you just ask.
  • A tiny CRM for a solo business

    Notes from your day become contacts, deals and follow-ups. Your own pipeline page shows them, and any AI can answer from the same rows.

    You say

    “Call with Sam at Acme, wants a quote for 3 pages by Friday.”

    Then
    • Sam goes into contacts, the quote into deals, and a follow-up for Friday.
    • A webhook posts the new deal to your Slack.
    • “Who am I waiting to quote?” gets an exact list from any AI.
  • One memory for your coding agents

    Claude Code, Codex and Cursor read and write decisions and known issues as rows, so no agent starts cold.

    You say

    “Idea for the budget app: a CSV export. Save it for later.”

    Then
    • You planned it on your phone in ChatGPT.
    • It is there when Claude Code starts at your desk, with the date and the chat it came from.

LaterApp makers: let your users connect their stash so your app knows them, with their permission.

What you get

Everything, on every plan.

Free, Pro and Builder have the same features. They differ only in limits: how many AIs, how much storage, and how many smart writes and reads.

  • One connector for every AI

    Claude, ChatGPT, Gemini, Cursor, Claude Code, Codex and any MCP client.

  • API keys

    Limited to the tables you choose, read or write.

  • Webhooks

    Your URL hears about new, changed and deleted rows, signed so you can check them.

  • Scheduled queries

    “Every Monday at 8am, send last week’s runs to this URL.” “On the 1st, write last month’s spending by category into monthly_totals.” No AI involved.

  • Rules

    When a row is saved, a formula writes the next one, like next week’s 62.5 kg bench press.

  • Live updates

    The app, every open tab and your own code hear about a change the moment it is saved.

  • Files on rows

    A receipt photo on the £42 expense, a PDF contract on the Acme deal. Files come from the Stashlet app, the desktop app, coding agents and the API.

  • Change history

    Every change keeps the old version and which AI or key made it.

  • Permissions and an access log

    Each AI and each key gets its own access, table by table, and every read is logged.

  • Import

    One prompt per AI, or upload the file of past chats that ChatGPT or Claude lets you download.

See the limits for each plan
Built for agents

How your agents talk to Stashlet.

Chat and coding agents use the MCP connector. Your code uses the REST API. Both reach the same tables, and the docs are written for your AI.

1The MCP connector

Add it once in Claude, ChatGPT, Gemini, Cursor, Claude Code, Codex or any MCP client. Sign-in is OAuth, and each AI gets its own permissions.

https://api.stashlet.app/mcpTools
describe_schema
Every table this AI can reach: what it holds, its fields and types, row count, common extra keys and two example rows. Takes tables, compact, limit and cursor.
create_table
Make a new table with typed fields. A connected AI must send a one-line description.
add_fields
Add fields to a table.
update_table
Change a table's description or its fields' descriptions.
save_rows
Save one row, or up to 100 at once.
update_row
Change some fields of a row.
delete_row
Delete a row. Its history stays.
query
Filter, sort, group and total one table.
search
Find rows by text across every table.
link_rows
Link two rows, like a photo and the people in it.
attach_file
Not available yet. Answers 501 not_available and attaches nothing.
read_history
Every past version of a row, and who changed it.
read_changes
Removed links, light switches and other changes history does not show.
decline_suggestion
Record that the person said no to a suggested table or field.
create_rule
Write a row from a formula whenever a matching row is saved.
list_rules
The rules whose tables this AI can read, and whether each runs.
update_rule
Change a rule's name, trigger or the row it writes.
delete_rule
Delete a rule. Rows it wrote stay.
read_guide
The full Stashlet guide, the same text as the skill. Reads nothing from the stash.

2REST for your code

Plain HTTPS and JSON from any language, with no SDK. Every tool has a matching endpoint. GET /v1/schema lists every table and field a key can reach, and every error says how to fix the call.

https://api.stashlet.app/v1

Each stash is its own isolated database. Your data isn’t stored alongside anyone else’s. Keys can be limited to the tables you choose.

3Docs your agent reads

  • agents.mdEvery tool and endpoint, with curl examples, errors and limits.
  • llms.txtThe standard index that points agents to the right docs.
Install prompt
Build with Stashlet: read https://stashlet.app/agents.md. My key is in STASHLET_KEY.

Paste it into Claude Code, Codex or Cursor.

The API

Plain HTTP. Your AI writes the rest.

The same row everywhere: Claude saved it, your code reads it, your webhook hears about it.

Save a row. What your code sends when you log something yourself.

curl https://api.stashlet.app/v1/tables/spending/rows \
  -H "Authorization: Bearer $STASHLET_KEY" \
  -H "Content-Type: application/json" \
  -d '{ "fields": { "amount": 42, "currency": "GBP", "place": "Tesco",
                   "category": "groceries", "date": "2026-11-17" } }'

{
  "id": "row_7kq2",
  "table": "spending",
  "fields": { "amount": 42, "currency": "GBP", "place": "Tesco", … },
  "by": { "type": "key", "name": "Budget app" },
  "created_at": "2026-11-17T09:41:00Z"
}

Query with totals. This month’s spending by category, biggest first.

curl https://api.stashlet.app/v1/query \
  -H "Authorization: Bearer $STASHLET_KEY" \
  -H "Content-Type: application/json" \
  -d '{ "table": "spending",
        "where": { "date": { "gte": "month_start" } },
        "group_by": ["category"],
        "totals": { "spent": { "sum": "amount" } },
        "sort": ["-spent"] }'

{
  "groups": [
    { "category": "groceries", "spent": 186.4 },
    { "category": "travel", "spent": 112 },
    { "category": "lunch", "spent": 61.5 }
  ],
  "totals": { "spent": 359.9 }
}

Install prompt. Paste into your coding agent. It reads the docs and writes the calls.

Build with Stashlet: read https://stashlet.app/agents.md. My key is in STASHLET_KEY.

Webhook. What your endpoint receives when Claude saves the row.

POST /hooks/stashlet HTTP/1.1
Content-Type: application/json
Stashlet-Delivery: dlv_91kd
Stashlet-Signature: t=1794908460,v1=5f3c…

{
  "event": "row.created",
  "table": "spending",
  "row": {
    "id": "row_7kq2",
    "fields": { "amount": 42, "currency": "GBP", "place": "Tesco",
                "category": "groceries", "date": "2026-11-17" }
  },
  "by": { "type": "ai", "name": "claude" },
  "at": "2026-11-17T09:41:00Z"
}
Pricing

Start free. The API is on every plan.

API keys, webhooks, scheduled queries, rules and live updates come with every plan. The plans differ only in limits, and Builder bills what you use beyond them.

  • Free

    Try it with the AI you use most.

    $0

    No card needed
    Create your free stash
    • 1 AI connected
    • 1 GB of storage
    • 1 GB of files
    • 1,000 smart writes a day
    • 100,000 reads a month
    • Change history kept 30 days
  • Pro

    Every AI you use, with room to grow.

    $12/month

    Billed monthly
    Choose Pro
    • Every AI you use
    • 5 GB of storage, then $1 per extra GB
    • 50 GB of files
    • 10,000 smart writes a day
    • 1,000,000 reads a month
    • Full change history
  • Builder

    For apps and scripts that save a lot.

    $29/month

    Billed monthly
    Choose Builder
    • Every AI you use
    • 5 GB of storage, then $1 per extra GB
    • 50 GB of files
    • 500,000 smart writes and 5,000,000 reads a month
    • Then $12 per million smart writes ($5 per million light ones) and $3 per million reads
    • Up to 50 write requests and 500 read requests a second
    • A batch of 100 rows counts as 100 smart writes
    • Full change history

Smart writes. More than a stored row: each one is checked for duplicates, indexed for search, versioned, synced live, and tracked for patterns that should become structure.

Light smart writes skip search and history, for high-volume data.

Every plan includes

  • API keys limited to chosen tables
  • Signed webhooks on every change
  • Scheduled queries to a URL, email or table
  • Rules that write the next row
  • Live updates over WebSocket
  • Permissions per AI and per key
  • Change history and a read log
  • Import by prompt or past-chat file

A smart write is one row saved or changed, so a batch of 100 rows counts as 100 smart writes. Your own use of the Stashlet app counts toward the totals too, but the app never stops you. A read is one get, query, search, history, describe-the-schema or change-log call. Builder’s speed limits count requests, not rows: up to 50 write requests and 500 read requests a second.

Prices in US dollars, plus tax where it applies. A yearly plan costs the same as 10 months.

Talk to any AI. Build only what you need.