Agent Behaviour — Cache Policy

🧠 Agent Behaviour: Cache Policy

This module defines how long agents should cache .llmfeed.json feeds, when to revalidate signatures, and how to handle offline scenarios.


🔄 Default Cache Duration (TTL)

Agents SHOULD respect a reasonable cache duration depending on the feed type and its use case:

Feed TypeRecommended TTL
`mcp`1 hour
`export`24 hours
`prompt`1 hour
`session`no caching (session state is volatile)
`credential`as indicated by the credential issuer
`pricing`15 minutes
  • If a feed declares a specific TTL (future extension), agents MUST respect it.

🔍 Signature Revalidation

Agents MUST revalidate signatures in the following cases:

  • When the cache TTL expires.
  • When the agent session restarts.
  • When a critical action is requested (such as invoking a capability based on the feed).
  • When an updated version of the feed is detected.

Partial or incremental revalidation may be supported in the future (e.g., via trust hints).


📡 Offline Mode

  • Agents MAY use cached, signed feeds in offline mode, provided the TTL is still valid.
  • If the TTL has expired and the feed cannot be revalidated, agents SHOULD:
    • Warn the user (if applicable).
    • Mark the feed as "stale".
    • Refrain from performing irreversible actions based on the stale feed.

This ensures that agents maintain proper context integrity and do not rely on outdated information.


🚦 Summary

The goal of this module is to ensure that .llmfeed.json feeds:

✅ Are used within their intended validity window.
✅ Are revalidated when needed to preserve trust.
✅ Are not blindly cached or reused without signature awareness.

This contributes to building a trustworthy, verifiable Agentic Web.

Ready to Implement? Get AI-Powered Guidance

Reading docs manually takes time. Your AI can digest the complete LLMFeed specification and provide implementation guidance tailored to your needs.

🎯

Quick Start

Essential concepts for immediate implementation

~22K tokens • 30s analysis • Core concepts
📚

Complete Mastery

Full specification with examples and edge cases

~140K tokens • 2min analysis • Everything
💡 Works with Claude, ChatGPT, Gemini⚡ Instant implementation guidance🎯 Tailored to your specific needs