oto.cx
airtable

airtable

airtable.

A row in an Airtable table: list, read, create, update, upsert, delete.

ajouter à otoauth : —prefix : airtable_*

setup & usage

⚠ airtable personal access token

create a personal access token at airtable.com/create/tokens, then paste it into oto (it starts with pat…).

  • two settings, NOT just one — this is airtable's #1 trap:

    • 1. scopes: data.records:read + data.records:write, data.recordComments:read + data.recordComments:write, schema.bases:read + schema.bases:write

    2. access: explicitly add each base (or the whole workspace) that the token must see

  • a token with all the scopes but no base granted answers 200 with an empty list, never an error: oto's "test connection" button detects it and says so
  • byo-only: no shared oto key — an airtable token is tied to an account and to specific bases
    • ▸ read and write in a base
  • "which bases can I reach?" → airtable_base() (returns the appXXXXXXXX — the starting point for everything else)
  • "what is in this base?" → airtable_table(base_id="app…") (tables, fields, types, options, views)
  • "which columns exactly, before writing?" → airtable_field(base_id="app…", table_id="tbl…")
  • "the rows where the status is Done" → airtable_record(base_id="app…", table="tbl…", filter_by_formula="{Status}='Done'")
  • "add these 40 prospects" → airtable_record(op="create", records=[{"Name": "…", "Email": "…"}, …]) (split into batches of 10 automatically)
  • "update if the email exists, create otherwise" → airtable_record(op="upsert", records=[…], merge_on=["Email"])
  • "fix this row" → airtable_record(op="update", record_id="rec…", fields={"Status": "Signed"})
  • "comment on this row" → airtable_comment(op="create", base_id=…, table=…, record_id=…, text="follow-up sent")
  • "attach this PDF" → airtable_attachment(base_id=…, record_id=…, field="Attachments", filename="quote.pdf", content_type="application/pdf", file_base64=…)
  • "create a tracking table" → airtable_table(op="create", name="Tracking", fields=[{"name": "Name", "type": "singleLineText"}, {"name": "Status", "type": "singleSelect", "options": {"choices": [{"name": "To do"}, {"name": "Done"}]}}])
    • · names vs identifiers, typecast, batches
  • use identifiers, not names: tbl…, fld… are stable; a table or column name changes as soon as someone renames it in the interface, and the automation silently breaks. airtable_table and airtable_field return the identifiers.
  • `typecast` is disabled by default, on purpose: at airtable it is not a convenience conversion but a schema change triggered by a data write — it creates the missing option of a select, even a record in the linked table of a *linked record* field. writing "Signed" into a select that only knows "Signee" therefore fails outright instead of adding a duplicate option. pass typecast=True when you WANT this behavior.
  • `replace=True` on an update is destructive: it is a PUT, every column not sent is CLEARED. the default (PATCH) only touches the columns sent.
  • batches: airtable refuses more than 10 rows per request and 5 requests/second per base. oto splits and spaces out automatically, up to 200 rows per call. if airtable cuts off because of the rate, the response carries aborted: "rate_limit" and the number of rows actually written — retry 30 seconds later with the rest.
  • long lists: airtable_record follows pagination up to max_records (100 by default) and sets more: true + offset if rows remain. nothing is silently truncated.
  • attachments > 5 MB: host the file and write its URL into the column — airtable_record(op="update", fields={"Attachments": [{"url": "https://…"}]}). airtable fetches it itself.
  • `airtable_sync` is not a CSV import: it feeds a table created in airtable via "Sync from other sources → API", and each push REPLACES the synced content. to load rows into an ordinary table, use airtable_record(op="create").
  • nothing can be deleted in the schema, and this is verified: no field DELETE (404), no table deletion, and a field PATCH carrying options is refused (422 "Changing a field's type or number precision is not currently supported"). in practice: a select option created by mistake with typecast=True can ONLY be removed in the Airtable interface. one more reason to leave typecast false.
  • deleting a base is impossible through the API: airtable_base(op="create") has no counterpart, it is done in the interface.
    • ⚙ multiple comptes

    this connector accepts multiple comptes: each stored credential becomes a named account (one name per compte), at your level, your team's or your org's.

  • the first one doesn't need a name; from the second on, each one carries its own (the first then takes the name "principal", renamable)
  • without specification, the agent takes the only stored compte, otherwise the one marked as default; otherwise the call is refused, naming the available comptes — never a random choice
  • target a compte for a call: _account="<name>" on the tool; list them: oto_identity(op='list', connector='airtable') (scope='org' or scope='group' for the org's or team's)
  • set the default: oto_identity(op='set', connector='airtable', identity_id='<name>'); rename: op='rename' with new_name
  • ⚠️ no tool walks through the comptes on its own: a total across several comptes is obtained by calling each one (_account) and adding up

    • outils

    chargement…

    utilisation

    claude code — installe le plugin oto — claude plugin marketplace add otomata-tech/oto-plugin — mcp + skill configurés.
    claude.ai — branche mcp.oto.cx/mcp en connecteur custom (guide)
    cli — pipx install oto-cli puis oto airtable …
    ← tous les connecteurscatalogue mcp →