Documentation Index
Fetch the complete documentation index at: https://docs.constants.io/llms.txt
Use this file to discover all available pages before exploring further.
Create Tools from Anything
Bring your code however you want — Constants handles the rest:- From GitHub
- From Upload
- From a Prompt
Typed Inputs & Outputs
Every tool gets a defined interface — typed input fields and structured outputs. This interface drives the web UI form, the API contract, and the MCP tool definition, keeping them all in sync:- Input types: Text, numbers, files, directories, arrays — with validation
- Output types: Text, numbers, files, JSON — rendered automatically in the UI
- Credentials: Declare what secrets a tool needs; they’re resolved and injected at runtime
- Compute tier: Choose resource allocation from small to GPU
Secure Sandbox Execution
Every tool runs in an isolated sandbox:- Fresh environment per execution — no state leaks between runs
- Resource tiers from lightweight scripts to GPU-powered ML inference
- Pre-installed packages including pandas, numpy, scipy, cloud SDKs, and ffmpeg
- Credential protection — secrets are encrypted and injected at runtime, never exposed to users or the sandbox directly
- File handling — upload inputs, download output artifacts
Run History & Artifacts
Every execution is recorded:| Field | Description |
|---|---|
| Inputs | The exact parameters and files provided |
| Status | Running, completed, failed, cancelled |
| Logs | Real-time streaming stdout/stderr |
| Timing | Start time and duration |
| Artifacts | Output files — downloadable from UI or API |
| Trigger | Who ran it (UI, Slack, API, agent) |
Built-in Utilities
Tools have access to pre-built Python utilities for common tasks:| Category | What You Get |
|---|---|
| Search & Scrape | Web search, website scraping |
| LLM | Call language models, extract structured data |
| Cloud Storage | GCP Storage, Firestore, BigQuery, Spanner, AWS S3 |
| Media | Image generation, speech-to-text, text-to-speech, video/audio processing |
| Files | File uploads, PDF processing |
| Data | LinkedIn data, YouTube downloads |
Run from Slack
Install the Constants bot in your Slack workspace and mention it to run tools:Agent Chat
Constants includes a built-in AI assistant that can:- Discover your tools and understand what they do
- Execute tools and return results
- Chain multiple tools into multi-step workflows
- Accept file attachments and route them to the right tool
REST API & MCP
Every tool is automatically available as:- REST API —
POST /v1/run/[toolName]for programmatic access - MCP endpoint — standard Model Context Protocol for AI agent integration
Teams & Organizations
Team Workspaces
Team Workspaces
Create team workspaces with shared tools, credentials, and usage quotas. Personal workspaces are auto-created for individual use.
Role-Based Access
Role-Based Access
Roles: owner, admin, member, viewer — enforced on every operation. Invite teammates by email.
Usage Quotas
Usage Quotas
Per-workspace limits on LLM, search, scrape, image, and media operations. Configurable per plan.
Shared Credentials
Shared Credentials
Security
| Layer | How It Works |
|---|---|
| Sandbox isolation | Each run in its own container with no access to other runs or host systems |
| Credential encryption | Secrets encrypted at rest and injected securely at runtime |
| API key scoping | Keys have configurable permission scopes with revocation support |
| Access control | Ownership, sharing, and role verification on every operation |
| Audit trail | Every run logged with full context — who triggered it, with what credentials, and what happened |
Conversation Mode
Refine your tools through natural language:- Request changes: “Make the filter case-insensitive”
- Debug failures: “Why did the last run fail?”
- Add features: “Add an option to output Excel format”
- Get explanations: “What does this tool do with the input file?”
What’s Next?
GitHub Integration
Import scripts and scan repos for tool candidates
API Reference
REST API, MCP, and Slack bot access