The two tools have clearly distinct purposes: 'recall' performs hybrid search for fragments by topic, while 'read_document' sequentially reads a full document. An agent can easily tell them apart without overlap.

'read_document' follows a clear verb_noun pattern, but 'recall' is a bare verb without an object, deviating slightly. Both are lowercase and action-oriented, so the set is mostly consistent with minor variation.

Two tools are borderline thin for a corpus server; typical well-scoped sets have 3–15 tools. While search and read cover a basic retrieval loop, additional tools for discovery or metadata could be expected.

The surface allows searching and reading documents, but there is no explicit way to list or browse available documents, nor to retrieve metadata such as dates or titles. This creates a notable gap for agents that need to discover what exists before reading.