interactiveai（interactiveai 搜索）

InteractiveAI

This skill helps you use InteractiveAI effectively across common workflows: instrumenting applications, migrating prompts, managing platform resources, and accessing documentation.

Core Principles

Follow these principles for ALL InteractiveAI work:

1. Documentation First: NEVER implement from memory. Always fetch current docs before writing code. See the documentation section below.
2. CLI and Schema Discovery First: Before changing InteractiveAI resources, discover the exact command shape with iai --help, iai <command> --help, and iai <command> <subcommand> --help. For structured prompts, use iai <type> schema when available.
3. Best Practices by Use Case: Check the relevant reference file below before implementing.
4. Default to Latest Prompt Fetches: When reading prompts or wiring prompt fetches, use the latest version unless the user explicitly asks for a different label or a fixed version.
5. Explicit Confirmation for Live Changes: Scaling services or replicas, or assigning the production label to a prompt version, changes live behavior. Do not do either unless the user explicitly confirms.
6. Use Latest InteractiveAI Versions: Unless the user specified otherwise or there is a good reason not to, use the latest InteractiveAI SDK and API versions.

Use case specific references

• instrumenting an existing function/application: references/instrumentation.md
• migrating prompts from a codebase into InteractiveAI: references/prompt-migration.md
• working with structured prompt types (routine, policy, variable, glossary, macro): references/structured-prompts.md
• further tips on using the iai CLI: references/cli.md

1. InteractiveAI CLI (iai)

Use the iai CLI to manage platform resources and inspect data. Unlike npx-based tools, iai must be installed locally.

Start by verifying availability and discovering the command shape:

iai --help
iai <command> --help
iai <command> <subcommand> --help

For structured prompt work, also inspect the type schema when available:

iai <type> schema

If iai is not installed, install the latest version:

go install github.com/Interactive-AI-Labs/interactive-cli/cmd/iai@latest
export PATH=$PATH:$(go env GOPATH)/bin

Credentials

Set these environment variables before using the SDK or CLI:

export INTERACTIVEAI_PUBLIC_KEY=pk-lf-...
export INTERACTIVEAI_SECRET_KEY=sk-lf-...
export INTERACTIVEAI_HOST=https://app.interactive.ai
export INTERACTIVE_API_KEY=$INTERACTIVEAI_PUBLIC_KEY:$INTERACTIVEAI_SECRET_KEY

INTERACTIVEAI_HOST must always be set. If the credentials are not available, ask the user for their API keys from InteractiveAI Settings.

Guardrails

• Ask for confirmation before running iai services or iai replicas commands that change capacity.
• Prompt versions receive latest automatically. Do not add --labels production unless the user explicitly confirms they want the new version live now.
• If the task would replace the version currently used in production, stop and ask for direct confirmation before proceeding.

Detailed CLI Reference

For common workflows, authentication details, and usage patterns, see references/cli.md.

2. InteractiveAI Documentation

Two methods to access InteractiveAI docs, in order of preference. Always prefer native web fetch/search tools over curl when available. The URL patterns below work with any fetch method; the curl examples are illustrative.

2a. Documentation Index (llms.txt)

Fetch the full documentation index:

curl -s https://docs.interactive.ai/llms.txt

This returns doc titles and relative paths. Prefix each path with https://docs.interactive.ai to form the full URL.

Alternatively, start from https://docs.interactive.ai and navigate to the relevant page.

2b. Fetch Individual Pages as Markdown

Any page listed in llms.txt can be fetched as markdown by appending .md to its path:

curl -s "https://docs.interactive.ai/sdk/tracing.md"
curl -s "https://docs.interactive.ai/sdk/prompts.md"
curl -s "https://docs.interactive.ai/cli/iai.md"

Documentation Workflow

1. Start with llms.txt to find the right page.
2. Fetch the specific markdown page once you know where the relevant guidance lives.