context7 icon indicating copy to clipboard operation
context7 copied to clipboard

Unable to retry, delete, or update a failed project import (Project Name Collision)

Open lmalma opened this issue 2 months ago • 3 comments

I encountered a workflow deadlock when trying to add a new library (documentation website). I initially tried to import a URL, which failed due to a "No snippets found" error (it seems it processed only the specific page and didn't crawl the tree).

When I attempted to create a new project with a corrected URL (appending /* to force crawling), I was blocked by a naming collision error because the internal project name is generated automatically based on the domain/path, and the "failed" project still occupies that name.

The core issues are:

  1. No "Delete" option: I cannot remove the failed project to start over.
  2. No "Update" option: I cannot change the target URL of the existing failed project to try again.
  3. No Custom Naming: I cannot manually name the project to avoid the collision.

Steps to Reproduce:

  1. Go to https://context7.com/add-library?tab=website.
  2. Enter the URL: https://docs.ituran.com/docs/ituran-api-v2/7009e73509778-ituran-documentation
  3. The process finishes with an error (see logs below) because it skipped the small files and didn't crawl subdirectories.
  4. Try to add a new library with the corrected URL: https://docs.ituran.com/docs/ituran-api-v2/*
  5. Error observed: The project /websites/ituran_ituran-api-v2 already exists.

Expected Behavior: I expect one of the following solutions:

  • A button to delete a project (especially one that failed initialization).
  • An option to overwrite an existing project when adding a new one with the same generated ID.
  • An option to edit the Source URL of an existing project and re-trigger the ingest process.

Logs / Context: Initial failure log:

Error: No (or too few) code snippets found in documentation files.
Repository cloned https://github.com/context7/ituran_ituran-api-v2 in 298ms
Checking for RST documentation...
Finding all markdown files in entire repository, no folders specified
No RST files found, skipping RST conversion
Found docs files 2 with total size 142
File too small, skipping: 7009e73509778-ituran-documentation.md (size: 74 characters)
File too small, skipping: README.md (size: 68 characters)
Finalizing project with error
Parsing completed
No snippets found in the repo
Process completed with failure

lmalma avatar Dec 01 '25 15:12 lmalma

Hey @lmalma, thanks for the heads up. I've deleted the old project with too few snippets. You can now reparse the website.

We've recently released a manage tab for github repositories, where you can manage all these configs yourself. And we're working on a feature for doing the same for websites. Until then, I hope this solves your problem.

fahreddinozcan avatar Dec 01 '25 15:12 fahreddinozcan

Thanks for clearing the previous project. I tried importing the library again using the wildcard URL: https://docs.ituran.com/docs/ituran-api-v2/*

However, the system still fails to ingest the documentation properly. Regarding the scraper:

  • Why is it failing to pick up the content?
  • Is there a specific convention or syntax I should use to ensure the scraper crawls all routes, sub-pages, and endpoints recursively?

Additional Issue: VS Code Extension API Key I also encountered a usability issue with the VS Code extension. When the popup for the API Key first appeared, I accidentally pressed Enter without pasting the key. Since then, I haven't been able to find any menu option, setting, or command in the Command Palette to input or update the API Key. How can I reset or enter the key manually?

lmalma avatar Dec 02 '25 00:12 lmalma

I've checked the website on my local, and it seems that there are public accessibility problems, and this is most likely to be the problem. Our crawler can't see the docs content, since it cannot authorize/login to any website. It seems that the crawler can't find any paths to crawl at the given url, so it halts. If this documentation belongs to you, please make sure it's publicly available without any authorization/firewall block by ip etc.

For your second question, MCP's api key configurations resides at mcp.json file at VS Code. To access this file, you should open the command palette (CMD+SHIFT+P at macos) and go to MCP: Open User Configuration, there you will see all your mcp clients and their configurations.

The expected json should be sth like below

VS Code Remote Server Connection

"mcp": {
  "servers": {
    "context7": {
      "type": "http",
      "url": "https://mcp.context7.com/mcp",
      "headers": {
        "CONTEXT7_API_KEY": "YOUR_API_KEY"
      }
    }
  }
}

VS Code Local Server Connection

"mcp": {
  "servers": {
    "context7": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "@upstash/context7-mcp", "--api-key", "YOUR_API_KEY"]
    }
  }
}

fahreddinozcan avatar Dec 02 '25 09:12 fahreddinozcan