MCP Tools Reference

Intelligence Engine exposes 15 MCP tools for code intelligence, search, analysis, and AI-powered documentation. All tools are prefixed ie_ and available via the intelligence-engine MCP server.

Prerequisite: Most tools require a project to be indexed first using ie_index.

Tool Categories

Category Tools
Indexing ie_index, ie_status
Search ie_query, ie_search_all
Graph Navigation ie_context, ie_cypher
Quality & Health ie_quality, ie_health
Change Analysis ie_detect_changes
AI Summaries ie_summarize, ie_batch_summarize
Documentation ie_wiki
Cross-Project ie_global_analysis
Memory ie_memory

1. ie_index

Index a project codebase for code intelligence. Parses source code (Python, JavaScript, TypeScript, Java, Go), builds a knowledge graph, and creates search indices. Must be run before using other ie_* tools on a project.

Supports incremental indexing: when mode='auto' (default), re-indexes only changed files if a previous index exists.

Parameters

Parameter Type Required Default Description
project_path string Yes - Path to project root directory
mode string No "auto" "auto" (incremental if possible), "full" (rebuild), "incremental" (fail if not possible)
force boolean No false Re-index even if already indexed

Example Usage 

{
  "project_path": "/home/simon/projects/paaf",
  "mode": "auto"
}

Example Output 

Indexed project 'paaf': 457 entities, 1823 relationships
Files: 42 Python files parsed
Time: 12.3s

2. ie_status

Check the index status of projects. Shows which projects are indexed, their entity counts, and when they were last indexed.

Parameters

Parameter Type Required Default Description
project string No null Project name. If omitted, shows all indexed projects.

Example Usage 

{}

{
  "project": "paaf"
}

Example Output 

Indexed Projects:
  paaf         457 entities   2026-03-05 14:30
  openclaw     312 entities   2026-03-04 09:15
  agent-comm    89 entities   2026-03-03 11:42

3. ie_query

Search a project’s codebase using hybrid BM25 + graph search. Returns ranked results with file:line citations and graph context (callers/callees).

Parameters

Parameter Type Required Default Description
project string Yes - Project name (as indexed)
query string Yes - Search query text (max 500 chars)
method string No "hybrid" "hybrid" (BM25+semantic+graph), "bm25" (keyword only), "semantic" (vector only), "graph" (BM25+graph, no semantic)
top_k integer No 10 Number of results to return

Example Usage 

{
  "project": "paaf",
  "query": "authentication middleware",
  "method": "hybrid",
  "top_k": 5
}

Example Output 

Results for "authentication middleware" in paaf (hybrid):

1. authenticate_request (function)        score: 0.92
   src/middleware/auth.py:45
   Callers: app_factory, register_routes
   Callees: verify_token, get_user_session

2. AuthMiddleware (class)                 score: 0.87
   src/middleware/auth.py:12
   Methods: __init__, __call__, authenticate_request

4. ie_search_all

Search across ALL indexed projects simultaneously. Returns results grouped by project with relevance scores. In shared DB mode, uses a single vector query (much faster).

Parameters

Parameter Type Required Default Description
query string Yes - Search query text (max 500 chars)
top_k integer No 10 Max results per project

Example Usage 

{
  "query": "database connection pool",
  "top_k": 5
}

Example Output 

Cross-project results for "database connection pool":

paaf (3 results):
  1. create_pool (function)     src/db/pool.py:22        score: 0.91
  2. DatabasePool (class)       src/db/pool.py:5         score: 0.85

openclaw (2 results):
  1. get_connection (function)  src/storage/db.ts:44     score: 0.78
  2. ConnectionManager (class)  src/storage/db.ts:10     score: 0.72

5. ie_context

Get 360-degree context for a code entity – callers, callees, and relationships. Shows who calls it, what it calls, and related entities.

Parameters

Parameter Type Required Default Description
project string Yes - Project name (as indexed)
entity_name string Yes - Entity name to look up (e.g. "scan_debt")
depth integer No 1 Depth for relationship traversal (1-3)

Example Usage 

{
  "project": "paaf",
  "entity_name": "scan_debt",
  "depth": 2
}

Example Output 

Entity: scan_debt (function)
File: src/analyzers/debt.py:89
Complexity: 12

Callers (who calls scan_debt):
  main                  src/cli.py:45
  run_analysis          src/runner.py:112

Callees (what scan_debt calls):
  parse_file            src/parsers/base.py:23
  calculate_score       src/analyzers/scoring.py:56
  format_results        src/output/formatter.py:78

Blast Radius: 5 entities affected by changes to scan_debt

6. ie_cypher

Execute a read-only Cypher query against a project’s knowledge graph (KuzuDB backend). Write operations (CREATE, DELETE, SET, DROP, etc.) are blocked.

CRITICAL: Use n.entity_type not n.type for filtering entity types. Always include a LIMIT clause to avoid overwhelming results.

Entity Table Columns

id, project, name, entity_type, file, line_start, line_end, is_exported, docstring, params, decorators, bases, code_snippet, language, complexity, summary, summary_provider, qa_history

Edge Types

CALLS, IMPORTS, EXTENDS, DEFINES, METHOD_OF

Parameters

Parameter Type Required Default Description
project string Yes - Project name (as indexed)
query string Yes - Cypher query string (max 2000 chars)

Example Usage 

{
  "project": "paaf",
  "query": "MATCH (n:Entity) WHERE n.entity_type = 'class' RETURN n.name, n.file LIMIT 20"
}

Example Output 

| n.name           | n.file                    |
|------------------|---------------------------|
| DebtScanner      | src/analyzers/debt.py     |
| AuthMiddleware    | src/middleware/auth.py     |
| DatabasePool     | src/db/pool.py            |

7. ie_quality

Get code quality metrics for a project: composite score, complexity distribution, documentation coverage, function length stats, module coupling, and fan-in/out metrics.

Parameters

Parameter Type Required Default Description
project string Yes - Project name (as indexed)

Example Usage 

{
  "project": "paaf"
}

Example Output 

Composite Score: 69.1/100

Complexity:
  Average: 1.6
  Max: 43 (main)
  High complexity (>10): 5 entities

Doc Coverage: 39.6% (181/457)

Function Length:
  Average: 12.3 lines
  Max: 156 lines (main)

Module Coupling:
  Average fan-out: 3.2
  Max fan-out: 18 (cli.py)
  Average fan-in: 2.1
  Max fan-in: 24 (utils.py)

8. ie_health

Get code health metrics from graph analysis: dead code count, circular dependencies, and hub concentration.

Parameters

Parameter Type Required Default Description
project string Yes - Project name (as indexed)

Example Usage 

{
  "project": "paaf"
}

Example Output 

Dead Code: 239 entities (no callers, not exported)
Cycles: 0 circular dependency chains
Hub Concentration: 12.0% (top 5 entities account for 12% of all edges)

Top Hubs:
  main              18 connections
  utils.format      14 connections
  db.get_connection  12 connections

Dead Code Samples:
  _legacy_parse     src/parsers/old.py:34
  unused_helper     src/utils/misc.py:89

9. ie_detect_changes

Analyze the blast radius of changed files to assess risk. Identifies entities in the changed files and traces their callers/dependents.

Parameters

Parameter Type Required Default Description
project string Yes - Project name (as indexed)
files string[] Yes - List of changed file paths (relative to project root)

Example Usage 

{
  "project": "paaf",
  "files": ["src/db/pool.py", "src/middleware/auth.py"]
}

Example Output 

Changed files: 2
Entities in changed files: 8
Directly affected callers: 14
Total blast radius: 22 entities

High-risk changes:
  DatabasePool.get_connection    12 callers affected
  authenticate_request            6 callers affected

Affected files:
  src/cli.py
  src/runner.py
  src/api/routes.py

10. ie_summarize

Generate an AI-powered natural language summary for a single code entity. Requires an LLM provider to be configured (Claude, OpenAI, Gemini, or Ollama).

Parameters

Parameter Type Required Default Description
project string Yes - Project name (as indexed)
entity_name string No null Entity name (alternative lookup)
entity_id string No null Full entity ID
provider string No null LLM provider (auto-detected if omitted)
max_tokens integer No 300 Max tokens for summary

Example Usage 

{
  "project": "paaf",
  "entity_name": "scan_debt"
}

Example Output 

scan_debt (function) - src/analyzers/debt.py:89

Scans a Python project directory for technical debt indicators. Walks the
file tree, parses each Python file using AST, and evaluates complexity,
documentation coverage, and code smells. Returns a DebtReport with per-file
scores and an aggregate project health score.

Called by: main, run_analysis
Calls: parse_file, calculate_score, format_results
Complexity: 12

11. ie_batch_summarize

Generate AI summaries for all entities in a project. Processes entities sequentially with a configurable delay between LLM calls for rate limit protection. Skips external and module nodes.

Parameters

Parameter Type Required Default Description
project string Yes - Project name (as indexed)
entity_types string[] No null Filter to specific types (e.g. ["function", "class"])
provider string No null LLM provider (auto-detected if omitted)
max_tokens integer No 300 Max tokens per summary
delay_seconds number No 1 Pause between LLM calls (rate limit protection)
skip_existing boolean No true Skip entities that already have summaries

Example Usage 

{
  "project": "paaf",
  "entity_types": ["function", "class"],
  "delay_seconds": 2,
  "skip_existing": true
}

Example Output 

Batch summarization for paaf:
  Total entities: 457
  Skipped (existing): 120
  Skipped (external/module): 89
  To summarize: 248

Progress: 248/248 complete
Errors: 0
Time: 8m 32s

12. ie_wiki

Generate markdown documentation from the knowledge graph. Can produce a full project overview, a single module breakdown, or a single entity page.

Parameters

Parameter Type Required Default Description
project string Yes - Project name (as indexed)
scope string No "full" "full" (project overview), "module" (single file), "function" (single entity)
entity_name string No null Entity or module name (required for module/function scope)

Example Usage 

{
  "project": "paaf",
  "scope": "module",
  "entity_name": "src/analyzers/debt.py"
}

Example Output 

# Module: src/analyzers/debt.py

## Overview
Technical debt analysis module containing the core scanning logic.

## Classes
- **DebtScanner** - Main scanner class for debt analysis

## Functions
- **scan_debt** - Scans project for technical debt indicators
- **calculate_score** - Computes debt score from raw metrics

## Dependencies
- Imports from: src/parsers/base, src/output/formatter
- Imported by: src/cli, src/runner

13. ie_global_analysis

Analyze the entire codebase across ALL projects in shared DB mode. Runs community detection and health analysis on the unified knowledge graph. Each community shows which projects its members come from.

Parameters

Parameter Type Required Default Description
include_clusters boolean No true Run community detection across all projects
include_health boolean No true Compute dead code, cycles, hubs across all projects

Example Usage 

{
  "include_clusters": true,
  "include_health": true
}

Example Output 

Global Analysis (3 projects, 858 entities):

Communities:
  Cluster 1: paaf (scan_debt, calculate_score), openclaw (analyze)
  Cluster 2: agent-comm (comm_send, comm_poll, comm_ack)
  Cluster 3: paaf (DatabasePool), openclaw (StorageManager)

Cross-Project Health:
  Total dead code: 312 entities
  Cross-project cycles: 0
  Shared patterns: 4 (logging, config, db, auth)

14. ie_memory

Unified memory browser aggregating data from two sources: KuzuDB (per-entity summaries and Q&A history) and Minna Memory (cross-session SQLite memories including session handovers, decisions, known issues).

Parameters

Parameter Type Required Default Description
project string Yes - Project name (as indexed)
entity_name string No null Filter by entity name
source string No null Filter by source: "kuzu" or "minna"
search string No null Full-text search query (Minna FTS)
attribute string No null Filter by attribute type
minna_scope string No null "project" = filter to current project only, null = show all
limit integer No 50 Max records to return (1-500)
include_stats boolean No true Include summary statistics

Example Usage 

{
  "project": "paaf",
  "search": "authentication",
  "source": "minna"
}

{
  "project": "paaf",
  "entity_name": "scan_debt",
  "source": "kuzu"
}

Example Output 

Memory for paaf (2 sources):

KuzuDB (3 records):
  scan_debt     summary    "Scans project for technical debt..."
  scan_debt     qa         Q: "What does scan_debt return?" A: "A DebtReport..."

Minna (2 records):
  paaf          decision   "Switched from SQLite to KuzuDB for graph storage"
  paaf          gotcha     "ie_cypher uses entity_type not type"

Stats: 5 total records (3 kuzu, 2 minna)

Cypher Cheat Sheet

The ie_cypher tool accepts read-only Cypher queries against the KuzuDB knowledge graph. Below are 10 common query patterns.

Remember: Always use n.entity_type (not n.type), and always include a LIMIT clause.

1. Find All Classes 

MATCH (n:Entity)
WHERE n.entity_type = 'class'
RETURN n.name, n.file, n.line_start
LIMIT 50

2. Functions by Complexity (Highest First) 

MATCH (n:Entity)
WHERE n.entity_type = 'function' AND n.complexity > 5
RETURN n.name, n.complexity, n.file
ORDER BY n.complexity DESC
LIMIT 20

3. Callers of a Function 

MATCH (caller:Entity)-[:CALLS]->(target:Entity)
WHERE target.name = 'authenticate_request'
RETURN caller.name, caller.file
LIMIT 20

4. Class Hierarchy (Inheritance) 

MATCH (child:Entity)-[:EXTENDS]->(parent:Entity)
RETURN child.name AS subclass, parent.name AS superclass, child.file
LIMIT 30

5. Dead Code (No Callers, Not Exported) 

MATCH (n:Entity)
WHERE n.entity_type = 'function'
  AND n.is_exported = false
  AND NOT EXISTS { MATCH (caller:Entity)-[:CALLS]->(n) }
RETURN n.name, n.file, n.line_start
LIMIT 50

6. Hub Entities (Most Connections) 

MATCH (n:Entity)-[r]-(m:Entity)
WITH n, count(r) AS connections
WHERE connections > 5
RETURN n.name, n.entity_type, connections, n.file
ORDER BY connections DESC
LIMIT 20

7. Imports Between Modules 

MATCH (a:Entity)-[:IMPORTS]->(b:Entity)
RETURN a.file AS importer, b.name AS imported, b.file AS from_file
LIMIT 30

8. Count Entities by Type 

MATCH (n:Entity)
RETURN n.entity_type AS type, count(n) AS count
ORDER BY count DESC
LIMIT 20

9. Functions in a Specific File 

MATCH (n:Entity)
WHERE n.file = 'src/analyzers/debt.py'
  AND n.entity_type = 'function'
RETURN n.name, n.line_start, n.complexity
ORDER BY n.line_start
LIMIT 50

10. Cross-Project Search (Shared DB Mode) 

MATCH (n:Entity)
WHERE n.name CONTAINS 'auth'
RETURN n.project, n.name, n.entity_type, n.file
ORDER BY n.project
LIMIT 30

Quick Reference

Tool One-liner Key params
ie_index Index a project for code intelligence project_path, mode
ie_status Check which projects are indexed project (optional)
ie_query Search within a single project project, query, method
ie_search_all Search across all projects query, top_k
ie_context Entity context: callers, callees, blast radius project, entity_name, depth
ie_cypher Raw Cypher queries on knowledge graph project, query
ie_quality Code quality composite score project
ie_health Dead code, cycles, hub analysis project
ie_detect_changes Blast radius of file changes project, files
ie_summarize AI summary for one entity project, entity_name
ie_batch_summarize AI summaries for all entities project, entity_types
ie_wiki Generate docs from graph project, scope
ie_global_analysis Cross-project architecture analysis include_clusters, include_health
ie_memory Unified KuzuDB + Minna memory browser project, search, source

This site uses Just the Docs, a documentation theme for Jekyll.