Tool Ecosystem and Categories

Relevant source files

• Qoder/prompt.txt
• README.md

Purpose and Scope

This page provides a comprehensive catalog of Qoder's tool ecosystem, documenting all available tools organized by functional category. Each tool enables Qoder to perform specific operations within the development environment, from code search and file manipulation to terminal execution and memory management.

For information about how these tools are executed in parallel or sequentially, see Parallel Tool Execution Strategy. For detailed rules about file editing tools specifically, see File Operations and Editing Rules. For validation tools usage, see Code Change Validation Pipeline.

Sources: Qoder/prompt.txt292-377

Tool Ecosystem Architecture

The following diagram illustrates how Qoder's 20+ tools are organized into eight functional categories, forming a comprehensive development assistant toolkit.

Sources: Qoder/prompt.txt292-339

Code Search and Analysis Tools

This category provides three complementary search mechanisms for navigating and understanding codebases.

Tool	Purpose	Primary Use Case
search_codebase	Symbol search or semantic search across codebase	Finding specific identifiers (classes, methods) or functionality descriptions
grep_code	Regular expression-based file content search	Pattern matching and exact text searches
search_file	File discovery using glob patterns	Locating files by name or extension pattern

Tool Selection Decision Flow

Symbol Search vs Semantic Search: Use symbol search when the query contains actual code identifiers like ClassName, methodName, or variableName. Use semantic search when describing functionality without specific symbol names. The decision rule: if the query contains PascalCase, camelCase, or "class/interface/method + Name" patterns, use symbol search.

Sources: Qoder/prompt.txt296-300 Qoder/prompt.txt346-350

File Operations Tools

File operations form the core of Qoder's code modification capabilities, with six tools spanning reading, creating, editing, and deleting files.

Tool Descriptions

Tool	Function	Constraints
list_dir	List contents of directory	Read-only operation
read_file	Read file contents with optional dependency viewing	Read-only operation
create_file	Create new files	Maximum 600 lines per file
search_replace	Make precise string replacements in existing files	Total line count across all replacements must stay under 600 lines; PRIMARY editing method
edit_file	Propose edits to existing files	FALLBACK ONLY - use only when search_replace is unsuitable
delete_file	Safely delete files	Destructive operation

Critical File Editing Rules

The system enforces strict penalties to ensure proper tool selection:

• MUST default to search_replace for editing files unless explicitly instructed to use edit_file, or face $100,000,000 penalty
• DO NOT try to replace entire file content with new content - this is very expensive, or face $100,000,000 penalty
• NEVER split short modifications (combined length under 600 lines) into several consecutive calls, or face $100,000,000 penalty
• MUST ensure original_text is uniquely identifiable in the file
• MUST match source text exactly including all whitespace and formatting
• NEVER allow identical source and target strings

Sources: Qoder/prompt.txt302-309 Qoder/prompt.txt236-243 Qoder/prompt.txt360-365

Terminal Operations Tools

Terminal operations enable command execution and output monitoring for build processes, test runs, and development servers.

Tool	Function	Usage Pattern
run_in_terminal	Execute shell commands	Use immediately when user requests execution; use background mode for long-running processes (servers, watch modes)
get_terminal_output	Get output from background terminal processes	Poll for output from commands started with background flag

Execution Constraints

Sequential Execution Requirement: run_in_terminal tools MUST ALWAYS be executed sequentially, never in parallel, to maintain proper execution order and avoid race conditions.

Sources: Qoder/prompt.txt312-314 Qoder/prompt.txt69 Qoder/prompt.txt75 Qoder/prompt.txt367-370

Code Validation Tools

The validation category contains a single mandatory tool that ensures code quality after every modification.

get_problems Tool

Function: Retrieve compile/lint errors in code files

Mandatory Usage Pattern:

1. After ANY code change, use get_problems to validate
2. Fix compilation/lint errors immediately
3. Continue validation until no issues remain
4. This applies to ALL changes, no matter how small

This tool is central to Qoder's validation pipeline and is documented extensively in Code Change Validation Pipeline.

Sources: Qoder/prompt.txt316-318 Qoder/prompt.txt146-151 Qoder/prompt.txt372-376

Task Management Tools

Task management tools enable Qoder to break down complex operations into trackable, verifiable steps.

Tool	Function	Usage Guidelines
add_tasks	Add new tasks to task list	Use for complex multi-step tasks (3+ distinct steps); use for non-trivial tasks requiring careful planning; skip for single straightforward tasks or trivial operations
update_tasks	Update task properties and status	Mark tasks complete ONLY after actual execution - NEVER before

Task Planning Principles

When using task management tools, follow these key principles:

• Break down complex tasks into smaller, verifiable steps
• Group related changes to the same file under one task
• Include verification tasks immediately after each implementation step
• Avoid grouping multiple implementations before verification
• Start with necessary preparation and setup tasks
• Group related tasks under meaningful headers
• End with integration testing and final verification steps

Sources: Qoder/prompt.txt320-323 Qoder/prompt.txt22-38 Qoder/prompt.txt245-250

Memory and Knowledge Tools

Memory tools enable Qoder to persist knowledge across sessions, addressing the context window limitation inherent to language models.

Tool	Function	Primary Use Case
update_memory	Store/update/delete knowledge and lessons learned	Persist project-specific information, user preferences, and experience lessons
search_memory	Search and retrieve codebase memory and knowledge	Query information across multiple knowledge documents for exploratory queries

Memory Categories and Scopes

Memory is organized into four categories with two scope levels:

Categories:

• user_prefer: Personal info, dialogue preferences, project-related preferences
• project_info: Technology stack, project configuration, environment setup
• project_specification: Development standards, architecture specs, design standards
• experience_lessons: Pain points to avoid, best practices, tool usage optimization

Scopes:

• workspace: Project-specific information
• global: Information applicable across all projects

When to Use Memory Tools

Use memory tools when:

• User explicitly asks to remember something
• Common pain points are discovered
• Project-specific configurations are learned
• Workflow optimizations are discovered
• Tool usage patterns that work well are identified

Search Memory Usage Guidelines

Use search_memory when:

• User asks questions requiring information across multiple knowledge documents
• Handling exploratory queries ("how to...", "what is...", "explain...")
• Analyzing code projects with insufficient existing context
• DO NOT use for simple tasks or when context is already sufficient

This memory system is documented in detail at Memory Management and Knowledge Systems.

Sources: Qoder/prompt.txt326-328 Qoder/prompt.txt152-174 Qoder/prompt.txt352-357

Web Operations Tools

Web operations enable real-time information retrieval and browser-based development workflows.

Tool	Function	Use Case
fetch_content	Fetch content from web pages	Retrieve documentation, API responses, or external resources
search_web	Search the web for real-time information	Access current information beyond training cutoff
run_preview	Set up preview browser for web servers	Enable live preview of web applications during development

Web Development Workflow

When building web applications:

1. Initialize project using CLI initialization tool
2. Start development server with run_in_terminal (background mode)
3. Use run_preview to set up preview browser
4. Before showing to user, use curl with run_in_terminal to access website and check for errors
5. Modern frameworks like Next.js have hot reload - development server keeps running in terminal

Sources: Qoder/prompt.txt330-334 Qoder/prompt.txt98-105

Rules and Guidelines Tools

The rules category provides access to detailed documentation about system rules and best practices.

Tool	Function	Usage
fetch_rules	Query detailed content of specific rules	Access comprehensive rule documentation when needed for task execution

Sources: Qoder/prompt.txt337-338

Tool Usage Philosophy and Selection

The following diagram illustrates the decision-making process for selecting appropriate tools based on query characteristics.

Core Philosophy

The tool usage philosophy emphasizes:

1. Check Required Parameters: Verify all required parameters are provided or can be reasonably inferred from context
2. Exact Value Matching: Use user-provided values EXACTLY as specified (especially quoted values)
3. No Optional Parameter Fabrication: DO NOT make up values for or ask about optional parameters
4. Analyze Descriptive Terms: Carefully analyze descriptive terms as they may indicate required parameter values

Priority Hierarchy

When multiple tools could accomplish a task:

1. File Editing: ALWAYS default to search_replace unless explicitly instructed to use edit_file
2. Terminal Operations: Execute commands immediately when user requests
3. Validation: MANDATORY use of get_problems after ALL code changes
4. Memory: Use only for exploratory queries or when context is insufficient

Sources: Qoder/prompt.txt340-377

Parallel vs Sequential Execution Rules

Qoder enforces strict rules about which tools can be executed in parallel versus sequentially to ensure correctness and performance.

Parallel Execution (Encouraged)

For maximum efficiency, execute independent operations simultaneously:

• Multiple read_file calls
• Multiple list_dir calls
• Multiple search_codebase calls
• Any combination of read-only tools

Performance Benefit: 3-5x faster execution compared to sequential calls

Example: When reading 3 files, run 3 tool calls in parallel to read all files into context simultaneously.

Sequential Execution (Mandatory)

The following tools MUST ALWAYS be executed sequentially:

1. run_in_terminal - Commands must run sequentially to ensure proper execution order and avoid race conditions
2. File editing tools (search_replace, edit_file, create_file) - File modifications must be sequential to maintain consistency

Violation Consequence: System instability and data corruption

Decision Matrix

Operation Type	Execution Mode	Rationale
Read-only tools	Parallel	Independent I/O operations, no state conflicts
File editing	Sequential	Maintains file consistency, prevents race conditions
Terminal commands	Sequential	Ensures proper execution order, prevents resource conflicts
Mixed read operations	Parallel	No dependencies between operations

Sources: Qoder/prompt.txt71-80 Qoder/prompt.txt68-69 Qoder/prompt.txt75

Tool Schema Compliance

All tools must be invoked following strict schema requirements:

1. ALWAYS follow the tool call schema exactly as specified
2. Provide all necessary parameters
3. NEVER call tools that are not explicitly provided in the available tools list
4. Use only the standard tool call format
5. The conversation may reference tools that are no longer available - verify tool availability before calling

Communication Guidelines

When discussing tool operations with users:

• NEVER refer to tool names directly to users
• Instead, describe what the tool is doing in natural language
• Focus on the action being performed rather than the technical implementation

Example:

• ❌ "I'll use the search_replace tool to modify the file"
• ✅ "I'll modify the file by replacing that code section"

Sources: Qoder/prompt.txt59-69