Tools

The Klever Connect MCP server exposes 6 tools (in MCP terminology: functions the assistant can call). The server is read-only — none of these tools mutate the knowledge base.

Tool list

Tool	Description
`query_context`	Search the knowledge base with structured filters and BM25-ranked full-text search.
`get_context`	Retrieve the FULL content of a single knowledge entry by ID.
`find_similar`	Find entries related to a known entry by comparing tags and type.
`get_knowledge_stats`	High-level overview: total entries, breakdown by type, top tags, categories.
`enhance_with_context`	Augment a natural-language question with the 5 most relevant entries, formatted as injectable context.
`search_documentation`	Natural-language documentation search with BM25 ranking, returns formatted markdown.

`query_context`

Search the knowledge base with structured filters and BM25-ranked full-text search. Filter by entry type, tag, or category.

Parameters

Name	Type	Required	Description
`query`	string	yes	Natural-language search query.
`type`	string	no	Filter by entry type (e.g. `code_example`, `best_practice`, `error_pattern`).
`tag`	string	no	Filter by a single tag.
`category`	string	no	Filter by category (e.g. `api-reference`, `examples`).
`limit`	number	no	Max results to return (default 10).

Example call

{
  "name": "query_context",
  "arguments": {
    "query": "sign transaction with web extension wallet",
    "type": "code_example",
    "limit": 5
  }
}

`get_context`

Retrieve the full content of a single knowledge entry by its ID.

Parameters

Name	Type	Required	Description
`id`	string	yes	The entry ID (returned by `query_context` or `search_documentation`).

Example call

{
  "name": "get_context",
  "arguments": { "id": "api-reference/connect-wallet/signTransaction" }
}

`find_similar`

Find entries related to a known entry by comparing tags and type. Useful for "show me more like this".

Parameters

Name	Type	Required	Description
`id`	string	yes	The reference entry ID.
`limit`	number	no	Max results (default 5).

Example call

{
  "name": "find_similar",
  "arguments": { "id": "api-reference/connect-wallet/signTransaction", "limit": 5 }
}

`get_knowledge_stats`

High-level overview of what's in the knowledge base. No parameters.

Returns an object with:

total — total entries
byType — counts grouped by entry type
byCategory — counts grouped by category
topTags — most-frequent tags

Example call

{
  "name": "get_knowledge_stats",
  "arguments": {}
}

Take a natural-language question and return the 5 most relevant knowledge entries, formatted as injectable context. Use this before asking the LLM to generate code — the response is meant to be pasted into the LLM's context window.

Parameters

Name	Type	Required	Description
`query`	string	yes	The user's natural-language question.

Example call

{
  "name": "enhance_with_context",
  "arguments": { "query": "How do I claim KFI rewards from a staking pool?" }
}

`search_documentation`

Natural-language documentation search with BM25 ranking. Returns formatted markdown — the most assistant-friendly tool for "show me how to do X".

Parameters

Name	Type	Required	Description
`query`	string	yes	Natural-language documentation query.
`limit`	number	no	Max results to return (default 5).

Example call

{
  "name": "search_documentation",
  "arguments": { "query": "how to send a KLV transfer with the connect SDK", "limit": 5 }
}

When to use which tool

Looking something up to read → search_documentation (markdown-formatted, optimized for the assistant to summarize).
Filtering by type/tag/category → query_context (structured filters).
Reading one entry in full → get_context (after you have the ID).
Discovering related entries → find_similar.
Pre-loading the LLM's context before code generation → enhance_with_context.
Auditing the knowledge base → get_knowledge_stats.

Tools

Tool list

`query_context`

`get_context`

`find_similar`

`get_knowledge_stats`

`enhance_with_context`

`search_documentation`

When to use which tool

Next steps