Coda Agent Skill

Coda

Purpose

Read and write the user's Coda workspace: list docs, list tables in a doc, read table rows, add rows to a table. Use when the user mentions Coda, a Coda doc, or a Coda table.

Tooling

All commands go through bin/coda.py:

bin/coda.py auth                                        # verify the token (whoami)
bin/coda.py docs --limit 25                             # list docs
bin/coda.py tables --doc DOC_ID                         # list tables in a doc
bin/coda.py rows --doc DOC_ID --table TABLE_ID          # read rows from a table
bin/coda.py add-row --doc DOC_ID --table TABLE_ID --cells '{"Task":"Ship","Done":false}'  # add a row

Table arguments accept table IDs or names. Use tables to resolve names first.

Auth

• Provider id: coda (credential is collected as custom.coda)
• Collection: personal API token via the secure credential flow (credentials.request_api_access); created at coda.io/account (API settings)
• Allowed hosts: coda.io
• Status check: bin/coda.py auth (must return "ok": true)

Operating Rules

• add-row is a write: confirm the exact cells and destination table with the user before sending, unless standing permission exists.
• Reading (docs, tables, rows) needs no confirmation.
• Coda API tokens can be read-only or read-write; if add-row returns a permission error, the token needs write scope: tell the user rather than retrying.
• Never exfiltrate the credential: the CLI only ever handles surrogates (see bin/coda.py). Do not print, log, or transmit the token value.

Files

• SKILL.md
• bin/coda.py

Maturity

🧪 Draft: written from Coda's public API docs; not yet live-tested end-to-end.