Configuration

Configure Tameshi behavior for CLI and VSCode integration.

VSCode Configuration

Configure Tameshi in VSCode Settings (Cmd+, or Ctrl+,) or .vscode/settings.json.

Essential Settings

{
  "tameshi.server.path": "/path/to/tameshi-lsp/target/release/tameshi-lsp",
  "tameshi.scan.onSave": "file",
  "tameshi.scan.include": ["**/*.sol"],
  "tameshi.scan.exclude": ["**/node_modules/**", "**/test/**"],
  "tameshi.llm.enabled": false
}

Server Configuration

Setting	Type	Default	Description
tameshi.server.path	string	-	Path to Tameshi LSP server binary
tameshi.server.args	string[]	[]	Additional LSP server arguments
tameshi.server.env	object	{}	Environment variables for server

Scan Behavior

Setting	Type	Default	Description
tameshi.scan.onSave	string	"file"	When to scan: "none", "file", "workspace"
tameshi.scan.onSave.debounce	number	1000	Debounce delay in milliseconds
tameshi.scan.include	string[]	["**/*.sol"]	Files to scan (glob patterns)
tameshi.scan.exclude	string[]	["**/node_modules/**"]	Files to exclude
tameshi.scan.onIdle.enabled	boolean	true	Scan when editor is idle
tameshi.scan.onIdle.idleSeconds	number	30	Idle threshold in seconds

LLM Configuration

Setting	Type	Default	Description
tameshi.llm.enabled	boolean	false	Enable LLM scanning
tameshi.llm.provider	string	"openai"	Provider: "openai" (only OpenAI supported currently)
tameshi.llm.apiKey	string	-	API key (use ${env:VAR_NAME} for env vars)
tameshi.llm.model	string	"gpt-4"	Model name
tameshi.llm.temperature	number	0.2	Response creativity (0.0-2.0)
tameshi.llm.maxTokens	number	4000	Maximum response tokens
tameshi.llm.confidenceThreshold	number	0.5	Minimum confidence (0.0-1.0)
tameshi.llm.timeoutSeconds	number	120	Request timeout
tameshi.llm.useIrScanning	boolean	false	Send ThalIR instead of source

Findings Display

Setting	Type	Default	Description
tameshi.findings.view.groupBy	string	"severity"	Group by: "severity", "file", "rule"
tameshi.findings.minCorrelationScore	number	0.5	Minimum correlation score (0.0-1.0)
tameshi.findings.mergeMode	string	"merged"	Display: "raw" or "merged"

Editor Features

Setting	Type	Default	Description
tameshi.editor.hovers.enabled	boolean	true	Show hover information
tameshi.editor.gutterIcons.enabled	boolean	true	Show gutter icons
tameshi.editor.codeLens.enabled	boolean	true	Show CodeLens summaries
tameshi.editor.statusBar.enabled	boolean	true	Show status bar item

CLI Configuration

Environment Variables

Set defaults via environment variables:

export TAMESHI_MIN_SEVERITY=medium
export TAMESHI_OUTPUT_FORMAT=json
export OPENAI_API_KEY=sk-...

LLM Configuration File

Create .tameshi/llm-config.yaml in your project:

provider:
  type: openai
  model: gpt-4
  api_key: ${OPENAI_API_KEY}

enabled_scanners:
  - reentrancy
  - access_control
  - integer_overflow

global:
  default_temperature: 0.2
  default_max_tokens: 4000
  retry_attempts: 3
  timeout_seconds: 60
  use_ir_scanning: false

Example Configurations

Development Workflow

Fast feedback with on-save file scanning:

{
  "tameshi.scan.onSave": "file",
  "tameshi.scan.onSave.debounce": 500,
  "tameshi.scan.include": ["contracts/**/*.sol"],
  "tameshi.scan.exclude": [
    "**/node_modules/**",
    "**/test/**",
    "**/mocks/**"
  ],
  "tameshi.findings.view.groupBy": "severity",
  "tameshi.editor.codeLens.enabled": true,
  "tameshi.llm.enabled": false
}

Security Audit

Comprehensive scanning with LLM:

{
  "tameshi.scan.onSave": "none",
  "tameshi.llm.enabled": true,
  "tameshi.llm.provider": "openai",
  "tameshi.llm.model": "gpt-4",
  "tameshi.llm.apiKey": "${env:OPENAI_API_KEY}",
  "tameshi.llm.useIrScanning": true,
  "tameshi.findings.minCorrelationScore": 0.7,
  "tameshi.findings.mergeMode": "merged"
}

Team Workspace

Shared settings for consistency:

{
  "tameshi.server.path": "${workspaceFolder}/../tameshi-lsp/target/release/tameshi-lsp",
  "tameshi.scan.onSave": "file",
  "tameshi.scan.include": ["src/**/*.sol"],
  "tameshi.scan.exclude": [
    "**/node_modules/**",
    "**/.deps/**",
    "**/artifacts/**"
  ],
  "tameshi.findings.view.groupBy": "file",
  "tameshi.editor.hovers.enabled": true,
  "tameshi.editor.gutterIcons.enabled": true
}

CI/CD Pipeline

Bash script with environment variables:

#!/bin/bash

export TAMESHI_MIN_SEVERITY=critical
export TAMESHI_OUTPUT_FORMAT=json

tameshi scan run -i src/ --format json > findings.json

# Fail if critical findings exist
if grep -q '"severity":"Critical"' findings.json; then
  echo "Critical vulnerabilities found"
  exit 1
fi

Workspace vs User Settings

User Settings (global, all projects):

• Server path
• Editor preferences
• UI customization

Workspace Settings (project-specific, committed to git):

• Scan patterns
• Excluded directories
• LLM configuration
• Minimum thresholds

Commit .vscode/settings.json for team consistency.

Smart AI Rescan Configuration

{
  "tameshi.scan.ai.smartRescan": "file",
  "tameshi.scan.ai.smartRescan.debounce": 3000,
  "tameshi.scan.ai.smartRescan.contextLines": 2,
  "tameshi.scan.ai.smartRescan.batchThreshold": 5,
  "tameshi.scan.ai.smartRescan.minSeverity": "medium"
}

Options:

• off - Disabled
• file - Rescan individual files as they change
• batch - Batch multiple changed files for efficiency

API Key Management

Environment Variable (Recommended)

# In ~/.bashrc or ~/.zshrc
export OPENAI_API_KEY=sk-...

Then reference in settings:

{
  "tameshi.llm.apiKey": "${env:OPENAI_API_KEY}"
}

Direct (Not Recommended)

Avoid committing API keys to git:

{
  "tameshi.llm.apiKey": "sk-..."
}

Add to .gitignore:

.vscode/settings.json

Configuration File

Store in .tameshi/llm-config.yaml (add to .gitignore):

provider:
  api_key: ${OPENAI_API_KEY}

Next Steps

• CLI Reference - Command-line options
• VSCode Guide - Using the extension
• SARIF Export - Reporting integration