trending/tradingview-mcp
1
0
Fork 0
mirror of https://github.com/tradesdontlie/tradingview-mcp.git synced 2026-10-02 05:24:58 +08:00
Code Issues Packages Projects Releases Wiki Activity

docs: reconcile tool count to 84 across all docs

Browse Source 
The registered MCP tool count had drifted across docs and the server
instructions header — CLAUDE.md said 81, README/RESEARCH/server.js said
78. Actual count is 84 (verified: grep -c "server.tool(" src/tools/*.js).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
tradesdontlie
2026-07-21 15:06:21 -04:00
co-authored by Claude Opus 4.8
parent 55534aab8c
commit d77ed4a2f5
4 changed files with 5 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files
CLAUDE.md
+1 -1
View File
@@ -1,6 +1,6 @@
# TradingView MCP — Claude Instructions
81 tools for reading and controlling a live TradingView Desktop chart via CDP (port 9222).
84 tools for reading and controlling a live TradingView Desktop chart via CDP (port 9222).
## Decision Tree — Which Tool When
README.md
+1 -1
View File
@@ -351,7 +351,7 @@ npm test
Claude Code ←→ MCP Server (stdio) ←→ CDP (port 9222) ←→ TradingView Desktop (Electron)
```
- **Transport**: MCP over stdio (78 tools) + CLI (`tv` command, 30 commands with 66 subcommands)
- **Transport**: MCP over stdio (84 tools) + CLI (`tv` command, 30 commands with 66 subcommands)
- **Connection**: Chrome DevTools Protocol on localhost:9222
- **Streaming**: Poll-and-diff loop with deduplication, JSONL output to stdout
- **No dependencies** beyond `@modelcontextprotocol/sdk` and `chrome-remote-interface`
RESEARCH.md
+2 -2
View File
@@ -22,7 +22,7 @@ How does an agent reason about data that may be stale by the time it responds? W
### 3. Tool Granularity
Should an agent have one `read_chart` tool or 78 granular tools? This project chose granularity — separate tools for quote, OHLCV, indicator values, pine lines, pine labels, pine tables, pine boxes, etc.
Should an agent have one `read_chart` tool or 84 granular tools? This project chose granularity — separate tools for quote, OHLCV, indicator values, pine lines, pine labels, pine tables, pine boxes, etc.
The tradeoff: granular tools give the agent precise control and small payloads, but require the agent to know which tool to call (solved via `CLAUDE.md` decision trees and MCP server instructions). A single coarse tool would be simpler but would waste context on unneeded data.
@@ -54,7 +54,7 @@ The most impactful design decision was making all tools return compact output by
### Tool Count Does Not Confuse the Agent
78 tools seems excessive, but with clear MCP server instructions and a `CLAUDE.md` decision tree, Claude consistently selects the right tools. The key is descriptive tool names and the instruction block — not reducing tool count.
84 tools seems excessive, but with clear MCP server instructions and a `CLAUDE.md` decision tree, Claude consistently selects the right tools. The key is descriptive tool names and the instruction block — not reducing tool count.
### Pine Script Development is the Strongest Use Case
src/server.js
+1 -1
View File
@@ -22,7 +22,7 @@ const server = new McpServer(
description: 'AI-assisted TradingView chart analysis and Pine Script development via Chrome DevTools Protocol',
},
{
instructions: `TradingView MCP — 78 tools for reading and controlling a live TradingView Desktop chart.
instructions: `TradingView MCP — 84 tools for reading and controlling a live TradingView Desktop chart.
TOOL SELECTION GUIDE — use this to pick the right tool: