Dope-context

What it is for

When you return to a complex repository, knowing the right question is only half the work. Dope-context offers code search, documentation search, and combined retrieval surfaces to help locate relevant implementation and explanations.

The intended benefit is a shorter path from a task description to useful source references. This is a retrieval projection: the source files remain the evidence, and retrieved passages do not become authoritative decisions or historical receipts.

How it fits the workspace

The canonical container entrypoint starts a FastMCP server. Compose connects it to Qdrant, configured provider credentials, and a read-only code mount. Search implementations are present for code, docs, and combined results.

Index and synchronization operations also exist. Normal retrieval profiles filter those mutation tools, keeping a read-focused investigation distinct from index maintenance. The useful distinction is between asking an index a question and changing what the index contains.

A concrete way to evaluate it

For an unfamiliar parser, begin with a precise question such as “Where is empty input handled?” Retrieve relevant code and adjacent tests, open the actual files, and compare the result with the current branch. Follow up with documentation search for the intended contract.

1. Confirm workspace and index scope.Do not assume a result describes the current checkout.
2. Retrieve a focused passage.Keep file references attached to the answer.
3. Inspect the source.Resolve contradictions against implementation and tests.

Setup and current limits

Search depends on usable indexes, Qdrant, and configured embedding access. The inspected path uses Voyage credentials/network access; unavailable embeddings produce errors. The website did not run indexing or retrieval and makes no relevance, latency, or quality benchmark claim.