AugmentClaude
Apache-2.0ETLBackend

Add Connector

Build or extend a PipesHub connector for external data sources and APIs.

Installation

  1. Make sure Claude is on your device and in your terminal.

    Skills load from ~/.claude/skills/ when Claude Code starts up — so you need it on your machine first. If you don't have it yet, install it once with the command below, then run claude in any terminal to verify.

    One-time setup
    npm i -g @anthropic-ai/claude-code

    Already have it? Skip ahead.

  2. Paste into Claude Code or into your terminal.

    This copies the whole skill folder into ~/.claude/skills/add-connector-pipeshub-ai/ — the SKILL.md plus any scripts, reference docs, or templates the skill ships with. Works for every skill.

    What this command does
    1. Makes a new, empty temporary folder.
    2. Downloads the skill's folder, .claude/skills/add-connector, from github.com/pipeshub-ai/pipeshub-ai (branch main), the GitHub repository it's listed from. You get the version that is there now, not a copy we keep.
    3. Copies that folder to ~/.claude/skills/add-connector-pipeshub-ai/.
    4. Deletes the temporary folder.

    Installing runs none of the skill's files; it only downloads and copies them. When you later use the skill, Claude follows its SKILL.md and may run scripts that come with it. Read them first.

    How we check skills, and what we don't · Report this skill

    Faster alternative (instruction-only skills)

    Skips the clone and grabs only the SKILL.md file. Don't use this if the skill ships Python scripts, reference markdowns, or asset templates — they won't be downloaded and the skill will fail when it tries to load them.

    Quick install (SKILL.md only)
    Sign up to copy
  3. Restart Claude Code.

    Quit and reopen Claude Code (or any other agent that loads from ~/.claude/skills/). New skills are picked up on startup.

  4. Just ask Claude.

    Skills auto-activate when your request matches the skill's description — no slash command needed. Trigger phrases live in the skill's own frontmatter; you can read them in the “What this skill does” section above.

Prefer to read the source first? Open on GitHub.

When Claude uses it

Build a new PipesHub connector, or extend an existing one, end to end, following CONNECTOR.md. Use when asked to add, create, build or port a connector or integration for an external source (Jira, Drive, Slack, a database, a SaaS API), or to add sync of a new object type to an existing connector.

What this skill does

Add a PipesHub connector

Follow CONNECTOR.md and the documents it links in docs/connectors/. This skill is the order of work and the points where you must stop; the documents hold the detail. Where they disagree with the code, the code is right: say so, and fix the document in the same PR.

1. Understand before you design

  1. Read CONNECTOR.md, then docs/connectors/01-high-level-design.md and 02-low-level-design.md.
  2. Pick the reference connector of the same shape from the table in docs/connectors/05-step-by-step.md and read its connector class, client and behaviour tests.
  3. Read the vendor's API documentation for everything listed under "Find out yourself" in docs/connectors/04-intake-questions.md. Note the documentation page for each fact.

2. Ask, once, with recommendations

Send the human one message with the questions in 04-intake-questions.md that the documentation did not answer, and the decisions in docs/connectors/03-decisions.md that need them. For each, give your recommendation and what you found, so they can confirm. Always ask for a test account and credentials, and say what cannot be verified without one.

Do not write code until the human has answered. If they cannot provide a test account, carry on with unit tests against fakes, and say in the PR that the connector was not run against the real source.

3. Write the design into the PR description

Before code: the "Connector intake" section, each decision from 03-decisions.md with its reason, and the permission mapping per object type from docs/connectors/06-permissions.md. Show it to the human if any part of the permission mapping errs towards over-sharing or hides content the source shows.

4. Build in phases

Work through docs/connectors/05-step-by-step.md phase by phase. Run each phase's check and do not start the next phase until it passes. While writing any API call, apply docs/connectors/07-errors-and-logging.md. While writing any record or group, apply 06-permissions.md.

Stop and ask the human when:

  • the source's sharing cannot be expressed without over-sharing;
  • a needed API scope may not be grantable, or needs a paid tier;
  • the live API contradicts its documentation;
  • you would need to change shared code under backend/python/app/connectors/core/, backend/python/app/models/ or the graph providers.

5. Finish

  1. Run the greps in 07-errors-and-logging.md on the new connector and its client, and read every hit.
  2. Run the checks in docs/connectors/08-testing.md: the contract tests, the connector's tests, the whole unit suite, ruff on the changed files.
  3. Go through CONNECTOR_CHECKLIST.md item by item, and copy it, ticked, into the PR description. For each unticked item, say why.
  4. Report to the human what was verified against the real source, what only against fakes, and what is known not to work.

Related skills