Subject: Incomplete Functionality of GitHub Integration: Missing Full-Text Access to Repository Files
Dear Google Support Team,
I am reaching out regarding an issue with the GitHub integration within the Gemini web interface.
Despite successfully connecting my GitHub account and correctly configuring access permissions for a specific repository (both via OAuth and the Google Integrations GitHub App), a fundamental functional issue occurs when trying to analyze repository contents using Gemini.
Issue Description
-
Symptom: After importing a repository, Gemini correctly reads the directory tree and lists all file paths/filenames. However, it cannot access or read the actual text content of the files within the repository.
-
Error: System attempts to fetch individual file contents (such as
.mddocumentation or plain text files) result in aPERMISSION_DENIEDstatus. -
Outcome: The language model states that it cannot read the file contents and asks the user to manually copy and paste the text or upload the files separately.
Questions for the Support & Product Team
-
Purpose of the Integration: What is the intended use case for the GitHub import feature in Gemini if the model can read the file structure after an import, but cannot access or analyze the actual file contents?
-
Missing Full-Text Scope: Is the failure to fetch file contents (
PERMISSION_DENIED) a technical bug in the permission handling between the GitHub App and Gemini, or is this an intentional system limitation? -
Resolution: Are there plans to update the GitHub integration so that imported repositories can be fully loaded into the model’s context for analysis without requiring manual workarounds?
I request a technical clarification on this behavior and information on when or if full-text repository analysis will function as expected.
System Context:
-
Repository Access: Authorized via Google Integrations
-
Affected File Types: Markdown (
.md), Plaintext (.txt), Code files
Best regards,