grepai-mcp-tools
Reference for all GrepAI MCP tools. Use this skill to understand available MCP tools and their parameters.
Install
npx skills add https://github.com/yoanbernabeu/grepai-skills --skill grepai-mcp-toolsSKILL.md
GrepAI MCP Tools Reference
This skill provides a complete reference for all tools available through GrepAI's MCP server.
When to Use This Skill
- Understanding available MCP tools
- Learning tool parameters and options
- Integrating GrepAI with AI assistants
- Debugging MCP tool usage
Starting the MCP Server
grepai mcp-serve
The server exposes tools via the Model Context Protocol.
Available Tools
1. grepai_search
Semantic code search using embeddings.
Parameters
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
query |
string | Yes | - | Search query describing what to find |
limit |
number | No | 10 | Maximum results to return |
compact |
boolean | No | false | Return compact output (no content) |
format |
string | No | "json" | Output format: "json" or "toon" (v0.26.0+) |
Example Request
{
"tool": "grepai_search",
"parameters": {
"query": "user authentication middleware",
"limit": 5,
"compact": true,
"format": "toon"
}
}
Response (Compact)
{
"q": "user authentication middleware",
"r": [
{"s": 0.92, "f": "src/auth/middleware.go", "l": "15-45"},
{"s": 0.85, "f": "src/auth/jwt.go", "l": "23-55"},
{"s": 0.78, "f": "src/handlers/auth.go", "l": "10-40"}
],
"t": 3
}
Response (Full)
{
"query": "user authentication middleware",
"results": [
{
"score": 0.92,
"file": "src/auth/middleware.go",
"start_line": 15,
"end_line": 45,
"content": "func AuthMiddleware() gin.HandlerFunc {\n ..."
}
],
"total": 3
}
2. grepai_trace_callers
Find all functions that call a specified symbol.
Parameters
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
symbol |
string | Yes | - | Function/method name to trace |
compact |
boolean | No | false | Return compact output (no context) |
format |
string | No | "json" | Output format: "json" or "toon" (v0.26.0+) |
Example Request
{
"tool": "grepai_trace_callers",
"parameters": {
"symbol": "Login",
"compact": true
}
}
Response (Compact)
{
"q": "Login",
"m": "callers",
"c": 3,
"r": [
{"f": "handlers/auth.go", "l": 42, "fn": "HandleAuth"},
{"f": "handlers/auth_test.go", "l": 15, "fn": "TestLoginSuccess"},
{"f": "cmd/main.go", "l": 88, "fn": "RunCLI"}
]
}
Response (Full)
{
"query": "Login",
"mode": "callers",
"count": 3,
"results": [
{
"file": "handlers/auth.go",
"line": 42,
"caller": "HandleAuth",
"context": "user.Login(ctx, credentials)"
}
]
}
3. grepai_trace_callees
Find all functions called by a specified symbol.
Parameters
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
symbol |
string | Yes | - | Function/method name to trace |
compact |
boolean | No | false | Return compact output (no context) |
format |
string | No | "json" | Output format: "json" or "toon" (v0.26.0+) |
Example Request
{
"tool": "grepai_trace_callees",
"parameters": {
"symbol": "ProcessOrder",
"compact": true
}
}
Response (Compact)
{
"q": "ProcessOrder",
"m": "callees",
"c": 4,
"r": [
{"f": "services/order.go", "l": 45, "fn": "validateOrder"},
{"f": "services/order.go", "l": 48, "fn": "calculateTotal"},
{"f": "services/order.go", "l": 51, "fn": "applyDiscount"},
{"f": "services/order.go", "l": 55, "fn": "sendConfirmation"}
]
}
4. grepai_trace_graph
Build a complete call graph starting from a symbol.
Parameters
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
symbol |
string | Yes | - | Root function for the graph |
depth |
number | No | 2 | Maximum recursion depth |
compact |
boolean | No | false | Return compact JSON format |
format |
string | No | "json" | Output format: "json" or "toon" (v0.26.0+) |
Example Request
{
"tool": "grepai_trace_graph",
"parameters": {
"symbol": "main",
"depth": 3,
"compact": true
}
}
Response (Compact)
{
"q": "main",
"d": 3,
"r": {
"n": "main",
"c": [
{
"n": "initialize",
"c": [
{"n": "loadConfig"},
{"n": "connectDB"}
]
},
{
"n": "startServer",
"c": [
{"n": "registerRoutes"}
]
}
]
},
"s": {"nodes": 6, "depth": 3}
}
Response (Full)
{
"query": "main",
"mode": "graph",
"depth": 3,
"root": {
"name": "main",
"file": "cmd/main.go",
"line": 10,
"children": [
{
"name": "initialize",
"file": "cmd/main.go",
"line": 15,
"children": [...]
}
]
},
"stats": {
"nodes": 6,
"max_depth": 3
}
}
5. grepai_index_status
Check the health and status of the code index.
Parameters
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
verbose |
boolean | No | false | Include detailed information |
format |
string | No | "json" | Output format: "json" or "toon" (v0.26.0+) |
Example Request
{
"tool": "grepai_index_status",
"parameters": {
"verbose": true
}
}
Response
{
"status": "healthy",
"project": "/path/to/project",
"embedder": {
"provider": "ollama",
"model": "nomic-embed-text",
"status": "connected"
},
"store": {
"backend": "gob",
"location": ".grepai/index.gob"
},
"index": {
"files": 245,
"chunks": 1234,
"last_updated": "2025-01-28T10:30:00Z"
},
"daemon": {
"running": true,
"pid": 12345
}
}
Compact Format Reference
When compact: true, responses use abbreviated keys:
| Full Key | Compact Key | Description |
|---|---|---|
query |
q |
Search query or symbol |
results |
r |
Results array |
total |
t |
Total count |
count |
c |
Count |
score |
s |
Similarity score |
file |
f |
File path |
line |
l |
Line number(s) |
mode |
m |
Trace mode |
depth |
d |
Graph depth |
name |
n |
Node name |
children |
c |
Child nodes |
stats |
s |
Statistics |
function |
fn |
Function name |
Token Efficiency
Compact mode reduces tokens significantly:
| Response Type | Full | Compact | Savings |
|---|---|---|---|
| Search (5 results) | ~800 | ~150 | 81% |
| Trace callers (10) | ~600 | ~120 | 80% |
| Trace graph (depth 3) | ~1200 | ~250 | 79% |
Error Responses
Index Not Found
{
"error": "Index not found. Run 'grepai watch' first.",
"code": "INDEX_NOT_FOUND"
}
Embedder Connection Failed
{
"error": "Cannot connect to embedding provider. Is Ollama running?",
"code": "EMBEDDER_UNAVAILABLE"
}
Symbol Not Found
{
"error": "Symbol 'FunctionName' not found in index.",
"code": "SYMBOL_NOT_FOUND"
}
Invalid Parameters
{
"error": "Parameter 'query' is required.",
"code": "INVALID_PARAMETERS"
}
Best Practices for AI Integration
- Use compact mode: Reduces token usage by ~80%
- Limit results: Request only what you need
- Check status first: Use
grepai_index_statusbefore searches - Handle errors: Check for error responses
- Combine tools: Search + trace for full understanding
MCP Protocol Details
Request Format
{
"jsonrpc": "2.0",
"id": 1,
"method": "tools/call",
"params": {
"name": "grepai_search",
"arguments": {
"query": "authentication",
"limit": 5,
"compact": true
}
}
}
Response Format
{
"jsonrpc": "2.0",
"id": 1,
"result": {
"content": [
{
"type": "text",
"text": "{\"q\":\"authentication\",\"r\":[...],\"t\":5}"
}
]
}
}
Output Format
MCP tools reference summary:
📚 GrepAI MCP Tools Reference
Tools available:
1. grepai_search
- Semantic code search
- Params: query*, limit, compact
2. grepai_trace_callers
- Find function callers
- Params: symbol*, compact
3. grepai_trace_callees
- Find function callees
- Params: symbol*, compact
4. grepai_trace_graph
- Build call graph
- Params: symbol*, depth, compact
5. grepai_index_status
- Check index health
- Params: verbose
* = required parameter
Compact mode: ~80% token reduction
Related skills
find-skillsvercel-labs3.6MHelps users discover and install agent skills when they ask questions like "how do I do X", "find a skill for X", "is there a skill that can...", or express interest in extending capabilities. This skill should be used when the user is looking for functionality that might exist as an installable skill.handoffmattpocock883KCompact the current conversation into a handoff document for another agent to pick up.microsoft-foundrymicrosoft618KBuild, deploy, evaluate, optimize, fine-tune, and manage Microsoft Foundry agents, models, and resources end to end. USE FOR: foundry, azd ai agent, azd provision/deploy, hosted agent scaffold/develop/run/deploy/troubleshoot, prompt agent create, create agent, update agent, add tool to agent, invoke agent, agent.yaml, agent insights, pull agent insights, evaluate agent, batch eval, continuous eval, continuous monitoring, agent CI/CD, optimize prompt, improve prompt, prompt optimizer, optimize agcavemanjuliusbrussee544KUltra-compressed communication mode that cuts output tokens while keeping technical accuracy. Levels: lite, full, ultra and the wenyan variants. Use for /caveman, "caveman mode", "talk like caveman", "be brief" or "less tokens".