booklib-mcp-guide

name	booklib-mcp-guide
version	1.0
license	MIT
tags	["meta","mcp","tools","agent-guide"]
description	Use when working with BookLib MCP tools. Reference for lookup, review, remember, verify, guard — parameters, workflows, and anti-patterns.

BookLib MCP Tool Guide

BookLib exposes 5 MCP tools to AI coding agents. Each tool has a single purpose. This guide covers when to call each one, what parameters to pass, and common workflows.

Tool Reference

Tool	When to call	Key params
lookup	Before working with unfamiliar APIs or post-training deps	query (required), file, limit, source
review	User asks for deep code review of a specific file	skill_name, file_path (both required)
remember	User says "remember this", "capture", or makes a team decision	title (required), content, type, tags
verify	After writing code that uses unfamiliar or new APIs	file_path (required)
guard	After writing code that touches architecture or API choices	file_path (required)

Decision Tree

Need current docs for a post-training library  -> lookup
Need a full expert review of a file            -> review
User wants to save a decision or insight       -> remember
Code uses unknown imports                      -> verify
Code might violate team rules                  -> guard

Tool Details

lookup

Searches across post-training corrections, team knowledge, and expert skills. Returns structured principles with source citations.

Parameters:

• query (string, required): What you need to know. Include the library name and task.
• file (string): Path to the file being worked on. Adds language and component context.
• limit (number): Maximum results. Default: 3.
• source (enum: all/skills/knowledge): Filter by source type.

Prioritization order: (1) post-training gap corrections, (2) team knowledge nodes, (3) expert skill principles.

review

Audits a file against a named skill's principles. Returns structured findings with line references and citations.

Parameters:

• skill_name (string, required): The skill to review against (e.g., "effective-kotlin", "clean-code-reviewer").
• file_path (string, required): Path to the file to review.

remember

Creates a searchable knowledge node. Automatically indexed and auto-linked to related nodes.

Parameters:

• title (string, required): Short descriptive title.
• content (string): Detailed description in markdown.
• type (enum: insight/decision/pattern/note/research): Node type. Default: "insight".
• tags (string): Comma-separated tags.
• links (string): Link targets as "target:edge-type" pairs.

verify

Checks if a file's imports are covered by BookLib's index. Flags unknown post-training APIs that may need current docs.

Parameters:

• file_path (string, required): Path to the source file.

Returns: list of unknown imports with suggested documentation URLs.

guard

Checks if code contradicts captured team decisions. Compares code patterns against decision nodes.

Parameters:

• file_path (string, required): Path to the source file.

Returns: list of contradictions with the violated decision and source.

Common Workflows

1. Gap detection and resolution

verify file.ts            -> finds unknown import "@stripe/stripe-js"
lookup "stripe js v5 api" -> returns current v5 docs
                          -> write correct code using current API

2. Team knowledge enforcement

write code                -> make architectural choice
guard file.ts             -> flags: contradicts "use PaymentIntents not Charges"
                          -> fix the violation, commit

3. Knowledge capture during work

discover a useful pattern -> remember "retry with exponential backoff" --type pattern
make a team decision      -> remember "use PaymentIntents not Charges" --type decision

Anti-Patterns

• Don't call lookup for standard patterns. React hooks, Python builtins, Go stdlib -- you already know these. Only call for project-specific or post-training knowledge.
• Don't call verify on stdlib imports. Standard library imports are always known. Only check third-party imports.
• Don't call remember for ephemeral notes. Only capture durable decisions, patterns, and insights that future sessions should know about.
• Don't call review without a specific skill. The skill_name parameter is required -- pick the most relevant one or use "clean-code-reviewer" as default.
• Don't over-call tools. One lookup per topic is enough. Don't repeat the same query hoping for different results.

CLI-Only Commands (Not MCP Tools)

These features are available via the booklib CLI but not exposed as MCP tools:

• booklib connect <path> -- connect a documentation source
• booklib connect github releases <repo> -- index GitHub changelogs
• booklib connect notion database <id> -- index Notion pages
• booklib link <from> <to> --type <edge> -- connect two knowledge nodes
• booklib save-state --goal "..." --next "..." -- save session for handoff
• booklib nodes list / booklib nodes show <id> -- list and inspect knowledge
• booklib gaps / booklib resolve-gaps -- detect and fix knowledge gaps
• booklib analyze -- show affected files and post-training APIs
• booklib doctor -- health check