letta-guide

Fetches and caches official Letta documentation via an ETag-verified Node.js helper script.

3.2k|385|Updated Oct 25, 2025
One-click install
npx skills add https://github.com/letta-ai/letta-code --skill letta-guide
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: letta-guide
Source: https://github.com/letta-ai/letta-code/tree/main/src/skills/builtin/letta-guide
Command: npx skills add https://github.com/letta-ai/letta-code --skill letta-guide

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

AI agents often answer questions about Letta from stale training data, inventing CLI flags, settings, pricing, or API details that no longer exist. This Skill routes every docs.letta.com lookup through a cached, ETag-checked fetch helper so answers about Letta products, APIs, and SDKs come from live documentation instead of memory.

Core Features & Use Cases

  • Verified docs retrieval: Fetches the docs.letta.com llms.txt index and individual pages, validating the content-MD5 ETag against the body before caching.
  • Heading outline generation: Builds a local outline of the docs index so the agent picks a page URL that actually exists instead of guessing paths.
  • Fallback and escalation rules: Defines when to use fetch_webpage as a last resort, when to check status.letta.com, and when to escalate to Discord or GitHub issues.
  • Use Case: A user asks how to configure a new model provider in Letta Code. The agent runs the helper to fetch the current configuration docs page, reads the relevant section, and answers with a cited, up-to-date URL rather than recalling outdated flags.

Quick Start

Ask the agent how to configure a Letta feature, such as setting up a model provider or channel, and it will fetch the current official docs before answering.

Frequently Asked Questions about letta-guide

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I look up current Letta documentation from an AI agent?

Run the fetch-letta-docs.mjs helper with Node.js to download the docs.letta.com llms.txt index, then pass a specific page URL with --docs-url to fetch that page. The helper verifies the ETag against the body before caching.

Why should I not use fetch_webpage on docs.letta.com?

The fetch_webpage tool may serve stale content from its upstream cache and guessing URLs can miss pages under different paths. The helper fetches the live index first and validates content via ETag, so answers reflect current docs.

What are the requirements to run the Letta docs fetch script?

The script needs Node.js 18 or newer for native fetch, or curl as a fallback transport. It also requires a writable temporary directory for its cache, which can be overridden with the --cache-dir flag.

What happens if docs.letta.com is unreachable?

The skill falls back to fetch_webpage with a fresh query parameter and discloses the content may be stale. If that also fails, it says the docs are unreachable, gives a best-effort answer marked as possibly out of date, and links to docs.letta.com.

When should Letta product questions not use the docs fetch route?

Questions about the currently running agent, such as its active model, tools, or memory contents, should be answered through self-inspection and the self-configuration skill rather than fetching documentation.