Conversation
Signed-off-by: Mouad BANI <mouad-mb@outlook.com>
There was a problem hiding this comment.
Pull request overview
Adds documentation for a new skill intended to guide/scaffold adding Snowflake connectors to CDP (crowd.dev), and updates the repository’s README to list and show the new skill in the project structure.
Changes:
- Added a new skill document at
lfx-cdp-snowflake-connectors/SKILL.mddescribing an end-to-end workflow for scaffolding a Snowflake connector. - Updated
README.mdto include/lfx-cdp-snowflake-connectorsin the skills table and in the example directory tree.
Reviewed changes
Copilot reviewed 2 out of 2 changed files in this pull request and generated 4 comments.
| File | Description |
|---|---|
| README.md | Documents the new skill in the skills summary table and project structure tree. |
| lfx-cdp-snowflake-connectors/SKILL.md | Introduces the detailed skill specification/workflow for scaffolding Snowflake connectors in crowd.dev/CDP. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
| --- | ||
| name: scaffold-snowflake-connector | ||
| description: Use when adding a new snowflake-connector data source to crowd.dev — a new platform or a new source within an existing platform that needs buildSourceQuery, transformer, activity types, migration, and all associated type registrations scaffolded. | ||
| --- | ||
|
|
||
| # Scaffold Snowflake Connector | ||
|
|
||
| ## Overview | ||
|
|
||
| This skill scaffolds all code required to add a new snowflake-connector data source to crowd.dev. It covers up to 11 touch points, enforces zero-assumption practices, and requires explicit user validation for every piece of business logic before writing any file to disk. | ||
|
|
||
| **One platform, one source per run.** Multiple sources within a platform are done one run at a time. |
There was a problem hiding this comment.
This SKILL.md is missing the standard metadata used elsewhere: allowed-tools: in the frontmatter and the copyright/SPDX/tool-mapping HTML comment block after the frontmatter (see lfx-intercom/SKILL.md:1-14). Add these to keep installation/tooling behavior and licensing headers consistent across skills.
| Use `ToolSearch` with query `"LFX BI Layer get_all_sources"` to check whether the BI Layer tools are available in the current session. | ||
|
|
||
| - **Tools found**: proceed directly to Step 1. | ||
| - **Tools not found**: prompt the user to authenticate: | ||
| > "The LFX BI Layer MCP is not connected. To enable automatic schema lookup, run `/mcp` and select **'claude.ai LFX BI Layer'** to authenticate. Once done, say 'continue' and I'll proceed. Or say 'skip' to use the manual query approach instead." | ||
|
|
||
| Wait for the user's response. If they authenticate: re-check with `ToolSearch` and proceed to Step 1. If they skip or authentication still fails: proceed directly to Step 2 for all tables. |
There was a problem hiding this comment.
The instructions reference a ToolSearch tool, but this repo’s documented Claude Code tool vocabulary (docs/tool-mapping.md) and other skills’ allowed-tools lists don’t include ToolSearch. Replace this with an approach that uses the allowed/documented tools (or explicitly document/whitelist the correct tool name) so the skill can be executed as written.
| Use `ToolSearch` with query `"LFX BI Layer get_all_sources"` to check whether the BI Layer tools are available in the current session. | |
| - **Tools found**: proceed directly to Step 1. | |
| - **Tools not found**: prompt the user to authenticate: | |
| > "The LFX BI Layer MCP is not connected. To enable automatic schema lookup, run `/mcp` and select **'claude.ai LFX BI Layer'** to authenticate. Once done, say 'continue' and I'll proceed. Or say 'skip' to use the manual query approach instead." | |
| Wait for the user's response. If they authenticate: re-check with `ToolSearch` and proceed to Step 1. If they skip or authentication still fails: proceed directly to Step 2 for all tables. | |
| Attempt to call `mcp__claude_ai_LFX_BI_Layer__get_all_sources` to check whether the BI Layer MCP is available in the current session. | |
| - **Call succeeds**: proceed directly to Step 1. | |
| - **Call fails because the MCP is unavailable or not authenticated**: prompt the user to authenticate: | |
| > "The LFX BI Layer MCP is not connected. To enable automatic schema lookup, run `/mcp` and select **'claude.ai LFX BI Layer'** to authenticate. Once done, say 'continue' and I'll proceed. Or say 'skip' to use the manual query approach instead." | |
| Wait for the user's response. If they authenticate: retry `mcp__claude_ai_LFX_BI_Layer__get_all_sources` and proceed to Step 1 if it succeeds. If they skip or the retry still fails: proceed directly to Step 2 for all tables. |
| | `/lfx-setup` | Environment setup — prerequisites, clone, install, env vars, dev server. Adapts to Angular or Go repos | Interactive guide | Bash, Read, Glob, Grep, AskUserQuestion | | ||
| | `/lfx-test-journey` | Combine branches from multiple repos into worktrees for journey testing | Interactive | Bash, Read, Write, Edit, Glob, Grep, AskUserQuestion | | ||
| | `/lfx-intercom` | Add or fix Intercom integration against the LFX canonical pattern — audits JWT setup, shutdown, Auth0 claim, app IDs, CSP | Audit + fix | Bash, Read, Write, Edit, Glob, Grep, AskUserQuestion | | ||
| | `/lfx-cdp-snowflake-connectors` | Streamlines adding a new Snowflake connector to CDP — requires knowledge of the source specs | Interactive and guided | Bash, Read, Write, Edit, Glob, Grep, AskUserQuestion, LFX BI Layer MCP | |
There was a problem hiding this comment.
The tools column lists “LFX BI Layer MCP”, but the README note says tool names follow Claude Code conventions and docs/tool-mapping.md only defines a small set (Bash/Read/Write/Edit/Glob/Grep/AskUserQuestion/Skill/WebFetch). Consider removing this from the tools column (or adding a documented/tool-mapping entry for MCP tools) to avoid implying it’s a standard tool name.
| | `/lfx-cdp-snowflake-connectors` | Streamlines adding a new Snowflake connector to CDP — requires knowledge of the source specs | Interactive and guided | Bash, Read, Write, Edit, Glob, Grep, AskUserQuestion, LFX BI Layer MCP | | |
| | `/lfx-cdp-snowflake-connectors` | Streamlines adding a new Snowflake connector to CDP — requires knowledge of the source specs | Interactive and guided | Bash, Read, Write, Edit, Glob, Grep, AskUserQuestion | |
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> Signed-off-by: Mouad BANI <mouad-mb@outlook.com>
Signed-off-by: Mouad BANI <mouad-mb@outlook.com>
|
@mbani01 Thanks for the PR. Can you please put some test results? |
@niravpatel27 sure, description updated ✅ |
This pull request updates the documentation to add information about a new skill for streamlining the addition of Snowflake connectors to CDP. The main changes are updates to the skill summary table and the project directory structure to include the new
/lfx-cdp-snowflake-connectorsskill.Documentation updates for new Snowflake connector skill:
/lfx-cdp-snowflake-connectorsto the skills summary table, describing its purpose and capabilities.lfx-cdp-snowflake-connectors/SKILL.mdfile.Problem
Adding a new Snowflake connector data source to CDP requires touching up to 11 files across multiple services — enums, migrations, transformers, query builders, and type registrations. On top of that, each connector starts from a spec: requirements gathered from other teams requesting the integration and passed to the CDP team to implement. This process is manual, error-prone, and poorly documented. Validating and confirming specs against actual Snowflake data is time-consuming and often delays progress.
Solution
Introduces the
lfx-cdp-snowflake-connectorsskill — a structured, AI-assisted scaffold that guides through the full connector implementation in one session.Key strengths:
PlatformTypeenum to migration SQL to transformer and query builder, nothing is skipped.Results