Description
Currently, codegraph init and the underlying MCP server only support tracking and indexing a single project root directory. In multi-repo or microservice architectures, we often have multiple peer directories (e.g., project-A, project-B, project-C) where project-A depends on code from B and C.
When working inside project-A, the AI agent cannot access the context of B and C unless we initialize CodeGraph in their common parent directory and open the entire parent folder in our IDE/Agent. However, this forces the agent to index all other unrelated peer directories and disrupts the workflow of focusing on a specific project.
Proposed Solution
It would be highly beneficial if CodeGraph could support configuring extra indexing paths. This could be achieved via:
- An option in the CLI, such as
codegraph init --include /path/to/project-B
- Or a project-level configuration file (e.g.,
.codegraph/config.json) where users can specify an array of extra paths to include:
{
"includePaths": [
"../project-B/src",
"../project-C/src"
]
}
Description
Currently,
codegraph initand the underlying MCP server only support tracking and indexing a single project root directory. In multi-repo or microservice architectures, we often have multiple peer directories (e.g.,project-A,project-B,project-C) whereproject-Adepends on code fromBandC.When working inside
project-A, the AI agent cannot access the context ofBandCunless we initialize CodeGraph in their common parent directory and open the entire parent folder in our IDE/Agent. However, this forces the agent to index all other unrelated peer directories and disrupts the workflow of focusing on a specific project.Proposed Solution
It would be highly beneficial if CodeGraph could support configuring extra indexing paths. This could be achieved via:
codegraph init --include /path/to/project-B.codegraph/config.json) where users can specify an array of extra paths to include:{ "includePaths": [ "../project-B/src", "../project-C/src" ] }