Agent Tools

Search Package Source Code (No indexing required!)

Ask your coding agent: “Use package search to find how error handling is implemented in the requests Python library” or “Search the numpy package for array manipulation examples”

Index Documentation, Repositories, Research Papers, or Datasets

Try indexing public documentation, a repository, research papers, or HuggingFace datasets:

“Index https://docs.trynia.ai/”
“Index https://github.com/browser-use/browser-use”
“Index https://arxiv.org/abs/2401.12345” (arXiv papers)
“Index https://huggingface.co/datasets/openai/gsm8k” (HuggingFace datasets)
“Use nia deep research to compare the best GraphRAG frameworks and then index the one with least latency.”

Monitor Progress and Explore

Check your indexed resources using manage_resource:

“List my resources” or “Check the status of my indexing jobs”
Visit app.trynia.ai to see all your indexed content

Available Tools

Universal Indexing

index - Index repos, docs, research papers, or local folders (auto-detects type)
auto_subscribe_dependencies - Index all dependencies from package.json, requirements.txt, and more

Universal Search

search - Semantic search across repos, docs, papers, and local folders
nia_package_search_hybrid - AI-powered semantic search in package source code

Navigation & Reading

nia_read - Read content from repo, docs, package, or local folder
nia_grep - Regex search in repo, docs, package, or local folder
nia_explore - Browse file structure (tree/ls actions)

Resource Management

manage_resource - List, status, rename, delete, or subscribe to resources

Research & Analysis

nia_research - AI research with 3 modes: quick, deep, oracle
nia_advisor - Analyze your code against documentation for recommendations

Context Sharing

context - Cross-agent context sharing with memory types (scratchpad, episodic, fact, procedural)

Complete Tool Reference

Package Search (No Indexing Required)

`nia_package_search_hybrid`

Semantic search in package source code with optional regex filtering. Search 3,000+ packages from PyPI, NPM, Crates.io, and Go modules without indexing.Parameters:

registry (str): One of “npm”, “py_pi”, “crates_io”, “golang_proxy”
package_name (str): Package name (for Go: “org/repo” format)
query (str): Natural language search query
pattern (str, optional): Regex filter to narrow results
version (str, optional): Specific package version

Example:

# Ask your agent:
"Search the requests package for how error handling works"
"Find authentication examples in the fastapi package"

Best for: Understanding implementation patterns, finding usage examples, exploring package architecture.

Universal Indexing Tools

`index`

Universal indexing tool that intelligently detects and indexes GitHub repositories, documentation, research papers, or local folders.Auto-Detection:

GitHub URLs (containing “github.com”) → Repository indexing
arXiv URLs (containing “arxiv.org”) → Research paper indexing
HuggingFace dataset URLs (containing “huggingface.co/datasets”) → Dataset indexing
Local folder path provided → Local folder indexing
All other URLs → Documentation indexing

Parameters:

url (str, optional): GitHub repository URL, documentation site URL, arXiv ID/URL, or HuggingFace dataset URL (not needed for local folders)
resource_type (str, optional): Override auto-detection - “repository”, “documentation”, “research_paper”, “huggingface_dataset”, or “local_folder”

Repository-specific:

branch (str, optional): Branch to index (defaults to main branch)

Documentation-specific:

url_patterns (List[str], optional): URL patterns to include in crawling
exclude_patterns (List[str], optional): URL patterns to exclude from crawling
only_main_content (bool, default=True): Extract main content only, skip nav/footer
check_llms_txt (bool, default=True): Check for /llms.txt file
llms_txt_strategy (str, default=“prefer”): Strategy for handling llms.txt - “prefer”, “only”, or “ignore”
wait_for (int, optional): Wait time in milliseconds for dynamic content to load
include_screenshot (bool, optional): Capture screenshots of pages
max_age (int, optional): Cache max age in seconds

Local folder-specific:

folder_path (str, optional): Absolute path to a folder on your machine (e.g., /Users/me/project)
folder_name (str, optional): Display name for the folder (defaults to folder basename)
files (List[dict], optional): List of {path, content} dicts (auto-read if folder_path provided)

Usage: Monitor progress with manage_resource action “status”Examples:

# Index a repository (auto-detected)
index("https://github.com/owner/repo", branch="main")

# Index documentation (auto-detected)
index("https://docs.example.com", url_patterns=["*/api/*"])

# Index a research paper from arXiv
index("https://arxiv.org/abs/2401.12345")

# Index a HuggingFace dataset (auto-detected)
index("https://huggingface.co/datasets/openai/gsm8k")

# Index a local folder
index(folder_path="/Users/me/notes", folder_name="My Notes")

`get_github_file_tree`

Get file and folder structure directly from GitHub API without indexing.Parameters:

repository (str, required): Repository identifier (owner/repo format)
branch (str, optional): Branch name (defaults to repository’s default branch)
include_paths (List[str], optional): Only show files in these paths
exclude_paths (List[str], optional): Hide files in these paths
file_extensions (List[str], optional): Only show these file types
exclude_extensions (List[str], optional): Hide these file types
show_full_paths (bool, default=False): Show full paths instead of tree structure

Best for: Quick repository exploration without indexing overhead.

`auto_subscribe_dependencies`

Automatically index all dependencies from your project’s manifest file. Parses the manifest, maps packages to documentation URLs, and subscribes to indexed docs.

Supported Manifests:

package.json (npm/Node.js)
requirements.txt (Python)
pyproject.toml (Python)
Cargo.toml (Rust)
go.mod (Go)
Gemfile (Ruby)

Parameters:

manifest_content (str, required): Raw content of the manifest file
manifest_type (str, optional): Auto-detected from content if not provided
include_dev_dependencies (bool, default=False): Include devDependencies
max_new_indexes (int, default=150): Maximum new documentation indexes to start
dry_run (bool, default=False): Preview mode - shows what would happen without subscribing

Example:

# Ask your agent:
"Read my package.json and auto-subscribe to all dependencies"
"Index all the libraries from my requirements.txt"
"Use auto_subscribe_dependencies on my Cargo.toml"

This feature indexes 10-100 dependencies at a time. Best suited for Pro & Startup plans due to indexing volume.

Universal Search

`search`

Semantic search across indexed repositories, documentation, research papers, and local folders. Omit sources for universal hybrid search across all indexed content.Parameters:

query (str, required): Natural language search question
repositories (List[str], optional): List of repositories to search (owner/repo format)
data_sources (List[str], optional): Documentation/research paper sources (UUID, display_name, URL, or arXiv URL)
local_folders (List[str], optional): Local folder IDs to search (private, searched separately)
category (str, optional): Filter local folder results by classification category
search_mode (str, default=“unified”): Search scope - “unified”, “repositories”, or “sources”
include_sources (bool, default=True): Include source code/document snippets

Example:

# Universal search across everything
"Search for how authentication is implemented"

# Target specific sources
"Search the react repo for useEffect implementation"

Best for: Searching across your indexed knowledge base with natural language.

Navigation & Reading Tools

These tools work across all source types: repositories, documentation, packages, local folders, and HuggingFace datasets. The source_type parameter determines which source to target.

`nia_read`

Read content from any source type.Parameters:

source_type (str, required): “repository”, “documentation”, “package”, “local_folder”, or “huggingface_dataset”
For repository: source_identifier (owner/repo:path/to/file)
For documentation: doc_source_id (UUID, display_name, or URL) + path
For package: registry, package_name, filename_sha256, start_line, end_line
For local_folder: local_folder_id + path
line_start / line_end (int, optional): Line range
max_tokens (int, optional): Maximum tokens to return

`nia_grep`

Regex search in any source type.Parameters:

source_type (str, required): “repository”, “documentation”, “package”, “local_folder”, or “huggingface_dataset”
pattern (str, required): Regex pattern to search for
For repository: repository (owner/repo format)
For documentation: doc_source_id (UUID, display_name, or URL)
For package: registry, package_name
For local_folder: local_folder_id
path (str, optional): Path prefix to limit search scope
case_sensitive (bool, default=False): Case-sensitive matching
context_lines (int, optional): Lines of context around matches
output_mode (str, default=“content”): “content”, “files_with_matches”, or “count”

`nia_explore`

Browse file structure of any source type.Parameters:

source_type (str, required): “repository”, “documentation”, “local_folder”, or “huggingface_dataset”
action (str, default=“tree”): “tree” for full structure, “ls” for directory listing
For repository: repository (owner/repo format)
For documentation: doc_source_id
For local_folder: local_folder_id
path (str, default=”/”): Path for ls action

Examples:

# Read from a repository
nia_read(source_type="repository", source_identifier="vercel/ai-sdk:src/index.ts")

# Grep in documentation
nia_grep(source_type="documentation", doc_source_id="nextjs-docs", pattern="getServerSideProps")

# Explore local folder structure
nia_explore(source_type="local_folder", local_folder_id="my-notes", action="tree")

Resource Management

`manage_resource`

Unified resource management tool for repositories, documentation, research papers, HuggingFace datasets, and local folders.Actions:

Action: “list”

List all indexed resources with their status.Parameters:

action (str, required): “list”
resource_type (str, optional): Filter by “repository”, “documentation”, “research_paper”, “huggingface_dataset”, “local_folder”, or None for all
query (str, optional): Text filter (matches repo/display_name/url/title/id)
limit (int, default=10): Maximum items to return per section (1-100)
offset (int, default=0): Pagination offset
view (str, default=“auto”): Output detail level - “auto”, “summary”, “compact”, “detailed”
show_all (bool, default=False): Ignore limit/offset and show all matches

Action: “status”

Check indexing progress and status of any resource.Parameters:

action (str, required): “status”
resource_type (str, required): “repository”, “documentation”, “research_paper”, “huggingface_dataset”, or “local_folder”
identifier (str, required): Repository name (owner/repo), documentation UUID/URL, paper identifier, or local folder ID

Action: “rename”

Rename a resource for better organization.Parameters:

action (str, required): “rename”
resource_type (str, required): “repository”, “documentation”, “research_paper”, “huggingface_dataset”, or “local_folder”
identifier (str, required): Resource identifier (owner/repo, UUID/URL, or local folder ID)
new_name (str, required): New display name (1-100 characters)

Action: “delete”

Remove an indexed resource.Parameters:

action (str, required): “delete”
resource_type (str, required): “repository”, “documentation”, “research_paper”, “huggingface_dataset”, or “local_folder”
identifier (str, required): Resource identifier (owner/repo, UUID/URL, or local folder ID)

Subscribe to a globally indexed public source. If the source is already indexed by others, you get instant access without re-indexing.Parameters:

action (str, required): “subscribe”
resource_type (str, required): “repository”, “documentation”, “research_paper”, or “huggingface_dataset”
identifier (str, required): URL of the source to subscribe to

Example:

manage_resource(action='subscribe', identifier='https://github.com/vercel/ai-sdk', resource_type='repository')

Identifier Formats:

For repositories: owner/repo format (e.g., “facebook/react”) or GitHub URL
For documentation: UUID preferred, also supports display name or URL
For research papers: UUID or arXiv abs/pdf URL
For HuggingFace datasets: UUID or dataset_id (e.g., “openai/gsm8k”, “squad”)
For local folders: UUID or display name

Research & Analysis Tools

`nia_research`

AI-powered research with three modes for different use cases.Modes:

Mode	Description	Best For
`quick`	Fast web search	Finding repos, docs, quick answers
`deep`	AI agent research	Comparisons, evaluations, analysis
`oracle`	Full autonomous research	Complex multi-step investigations

Parameters:

query (str, required): Research question
mode (str, default=“quick”): “quick”, “deep”, or “oracle”

Examples:

# Quick web search
"Use nia_research quick to find the best GraphRAG frameworks"

# Deep analysis
"Use nia_research deep to compare Prisma vs Drizzle"

# Full autonomous research
"Use nia_research oracle to investigate authentication best practices for Next.js"

`nia_advisor`

Analyze your code against documentation for tailored recommendations. Compares your implementation with best practices from indexed docs.Parameters:

code (str, required): Your code to analyze
doc_source_id (str, required): Documentation source to compare against
question (str, optional): Specific aspect to focus on

Example:

# Ask your agent:
"Use nia_advisor to check my authentication code against the BetterAuth docs"
"Analyze my React component against the Next.js App Router documentation"

Best for: Getting documentation-grounded recommendations for improving your code.

Context Sharing

`context`

Cross-agent context sharing for saving, retrieving, and managing conversation contexts.Memory Types:

Type	Duration	Best For
`scratchpad`	1 hour	Temporary working memory
`episodic`	7 days	Session-based context
`fact`	Permanent	Learned facts about the codebase
`procedural`	Permanent	How-to knowledge and procedures

Actions:

Action: “save”

Save conversation context.Parameters:

action (str, required): “save”
title (str, required): Descriptive title
summary (str, required): Brief summary
content (str, required): Full conversation context
agent_source (str, required): Which agent created this (e.g., “cursor”, “claude-code”)
memory_type (str, optional): “scratchpad”, “episodic”, “fact”, or “procedural”
tags (List[str], optional): Searchable tags
edited_files (List[dict], optional): Files modified during conversation

Action: “list”

List saved contexts with optional filtering.

Action: “retrieve”

Get full context by ID.

Action: “search” / “semantic-search” / “keyword-search”

Search contexts using different strategies.

Action: “update”

Update existing context.

Action: “delete”

Delete a context.Example:

# Save context for later
"Save this conversation about authentication setup"

# Retrieve relevant context
"Search my saved contexts for database migrations"

Need Help? Join our community or reach out through app.trynia.ai for support and updates.

Getting Started

Features

Plugins and Rules

Examples & Guides

Enterprise

Privacy

Available Tools

Universal Indexing

Universal Search

Navigation & Reading

Resource Management

Research & Analysis

Context Sharing

Complete Tool Reference

`nia_package_search_hybrid`

`index`

`get_github_file_tree`

`auto_subscribe_dependencies`

`search`

`nia_read`

`nia_grep`

`nia_explore`

`manage_resource`

Action: “list”

Action: “status”

Action: “rename”

Action: “delete”

`nia_research`

`nia_advisor`

Getting Started

Features

Plugins and Rules

Examples & Guides

Enterprise

Privacy

​Available Tools

Universal Indexing

Universal Search

Navigation & Reading

Resource Management

Research & Analysis

Context Sharing

​Complete Tool Reference

​nia_package_search_hybrid

​index

​get_github_file_tree

​auto_subscribe_dependencies

​search

​nia_read

​nia_grep

​nia_explore

​manage_resource

​Action: “list”

​Action: “status”

​Action: “rename”

​Action: “delete”

​Action: “subscribe”

​nia_research

​nia_advisor

​context

​Action: “save”

​Action: “list”

​Action: “retrieve”

​Action: “search” / “semantic-search” / “keyword-search”

​Action: “update”

​Action: “delete”

Available Tools

Complete Tool Reference

`nia_package_search_hybrid`

`index`

`get_github_file_tree`

`auto_subscribe_dependencies`

`search`

`nia_read`

`nia_grep`

`nia_explore`

`manage_resource`

Action: “list”

Action: “status”

Action: “rename”

Action: “delete”

Action: “subscribe”

`nia_research`

`nia_advisor`

`context`

Action: “save”

Action: “list”

Action: “retrieve”

Action: “search” / “semantic-search” / “keyword-search”

Action: “update”

Action: “delete”