Anyway Traces

Adds observability and tracing to AI/LLM applications using the Anyway SDK for monitoring calls to providers like OpenAI and Anthropic.

Security Scan

The skill's description, instructions, and behavior are consistent with an observability/tracing helper for LLM apps, but the package has no source/homepage declared and does not list the environment variables or credentials the SDK will require — so verify the origin and telemetry behavior before installing.

Purpose & Capability

The name/description claim to add tracing via the Anyway SDK and the SKILL.md contains links and instructions about installing and initializing the Anyway SDK and configuring tracing. There are no unrelated binaries, credentials, or install steps in the metadata that contradict the stated purpose.

Instruction Scope

The instructions direct the agent to fetch documentation from docs.anyway.sh and to instrument applications (init SDK, call Traceloop.init/initialize, etc.). They do not instruct reading arbitrary local files or exfiltrating system secrets, but they do imply the SDK will auto-instrument LLM provider calls and send telemetry to the collector endpoint (https://collector.anyway.sh/). That means traces (which can include prompts, responses, or metadata) may be sent to a third party — users should confirm what data is collected and redacted.

Install Mechanism

This is an instruction-only skill with no install spec and no code files, which minimizes direct install risk. The SKILL.md links to upstream install docs and package names but does not itself download or run code.

Credentials

The skill metadata declares no required env vars, but the documentation content repeatedly advises using environment variables and initializing the SDK and references a collector endpoint. That mismatch (no declared telemetry/API token variables) is an omission: the SDK will likely require API keys or a collector token and will send telemetry, so the skill should have declared what credentials/config are needed.

Persistence & Privilege

always is false and there are no install scripts or config-path requests. The skill does not request permanent agent presence or elevated privileges.

Assessment

This skill appears to simply teach an agent how to use the Anyway observability SDK, which is coherent for a tracing helper — but take these precautions before using: - Verify the origin: there is no homepage or source declared. Confirm the publisher and that docs.anyway.sh is the official site for the SDK before trusting it. - Check telemetry behavior: the SKILL.md lists a collector endpoint (https://collector.anyway.sh/). Ask or read the SDK docs to learn exactly what is sent (full prompts, responses, API keys, headers, environment values) and whether sensitive fields are redacted by default. - Confirm credentials and env vars: the skill metadata lists none but the SDK likely needs tokens or configuration. Find the official SDK docs to discover required env vars (collector token, API keys) before providing secrets. - Review upstream packages: if you install anyway-sdk or @anyway-sh/node-server-sdk, inspect the package contents and repository (PyPI/GitHub/npm) to ensure they match the official project and to audit what network endpoints the SDK contacts. - Test in isolation: instrument a non-sensitive test app or sandbox to observe what telemetry is emitted and whether you can point the SDK at a self-hosted collector or disable telemetry. If you can provide the official repository or package names and the list of env vars the SDK requires, I can reassess and raise confidence.

Documentation Index

Fetch the complete documentation index at: https://docs.anyway.sh/llms.txt Use this file to discover all available pages before exploring further.

AI Agent Skill

Copy the Anyway skill to teach your AI coding agent how to instrument your applications

<Card title="Download SKILL.md" icon="download" href="https://docs.anyway.sh/skills.md"> Download the skill file and add it to your AI coding agent's context. </Card>

What is this?

A skill file is a markdown file that gives AI coding agents (Claude Code, Cursor, Windsurf, etc.) domain-specific knowledge. The Anyway skill teaches your agent how to add observability and tracing to your AI applications using the Anyway SDK.

With this skill installed, you can give your agent instructions like:

• "Add Anyway tracing to my app"
• "Instrument this function with workflow and task spans"
• "Set up the Anyway SDK with environment variables"

Skill

Copy the content below or use the download link above to add the skill to your AI coding agent's context.

---
name: anyway
description: Instrument AI/LLM applications with Anyway observability SDK. Use when adding tracing, monitoring, or observability to code that calls OpenAI, Anthropic, or other LLM providers.
user-invocable: false
---

Anyway is a financial OS for AI agents that lets you monetize agent outcomes and capture real value. Anyway SDK provides AI observability, cost tracking, and better monetization for LLM applications.

## SDKs

- **Python** (`anyway-sdk`): Python 3.10+
- **JavaScript** (`@anyway-sh/node-server-sdk`): Node.js 18+

## Documentation

Fetch the relevant page(s) below for up-to-date installation, configuration, and usage instructions:

- Quickstart: https://docs.anyway.sh/quickstart
- Python SDK installation: https://docs.anyway.sh/sdk/python/installation
- Python SDK configuration: https://docs.anyway.sh/sdk/python/configuration
- Python SDK tracing: https://docs.anyway.sh/sdk/python/tracing
- JavaScript SDK installation: https://docs.anyway.sh/sdk/js/installation
- JavaScript SDK configuration: https://docs.anyway.sh/sdk/js/configuration
- JavaScript SDK tracing: https://docs.anyway.sh/sdk/js/tracing

## Rules

- Never hardcode API keys — use environment variables or a secrets manager
- Always call `Traceloop.init()` (Python) or `initialize()` (JS) before any LLM calls
- For ESM/Next.js projects, always pass provider modules via `instrumentModules`
- The collector endpoint is `https://collector.anyway.sh/`
- The SDK auto-instruments calls to OpenAI, Anthropic, Cohere, Bedrock, Vertex AI, and other providers — manual span creation is only needed for workflow/task structure

Built with Mintlify.