Hermes BibleUnofficial Docs
← All docs Docs / Developer Guide

Context Compression and Caching

Context Compression and Caching Hermes Agent uses a dual compression system and Anthropic prompt caching to manage context window usage efficiently across long

Developer Guide
Section
Developer Guide
Source
hermesbible.com/docs/developer-guide/context-compression-and-caching

Context Compression and Caching Hermes Agent uses a dual compression system and Anthropic prompt caching to manage context window usage efficiently across long

Excerpt from the official Hermes Agent documentation, quoted for reference. View source

What this page covers

Section outline mirrored from the official Hermes Agent documentation. Follow any heading to read the complete text on the source site.

Upstream outline

  1. Pluggable Context Engine
  2. Dual Compression System
  3. 1. Gateway Session Hygiene (85% threshold)
  4. 2. Agent ContextCompressor (50% threshold, configurable)
  5. Configuration
  6. Parameter Details
  7. Codex gpt-5.5 threshold autoraise
  8. Computed Values (for a 200K context model at defaults)
  9. Compression Algorithm
  10. Phase 1: Prune Old Tool Results (cheap, no LLM call)
  11. Phase 2: Determine Boundaries
  12. Phase 3: Generate Structured Summary
  13. Goal
  14. Constraints & Preferences

Section map

Pluggable Context Engine

Maps pluggable context engine to the Developer Guide documentation path, with the source page reserved for exact commands and updates.

Dual Compression System

Maps dual compression system to the Developer Guide documentation path, with the source page reserved for exact commands and updates.

1. Gateway Session Hygiene (85% threshold)

Maps 1. gateway session hygiene (85% threshold) to the Developer Guide documentation path, with the source page reserved for exact commands and updates.

2. Agent ContextCompressor (50% threshold, configurable)

Maps 2. agent contextcompressor (50% threshold, configurable) to the Developer Guide documentation path, with the source page reserved for exact commands and updates.

Configuration

Maps configuration to the Developer Guide documentation path, with the source page reserved for exact commands and updates.

Parameter Details

Maps parameter details to the Developer Guide documentation path, with the source page reserved for exact commands and updates.

Codex gpt-5.5 threshold autoraise

Maps codex gpt-5.5 threshold autoraise to the Developer Guide documentation path, with the source page reserved for exact commands and updates.

Computed Values (for a 200K context model at defaults)

Maps computed values (for a 200k context model at defaults) to the Developer Guide documentation path, with the source page reserved for exact commands and updates.

Implementation notes

  • Use this Developer Guide doc as a navigation page first: the local clone mirrors the source structure and links each heading back to the authoritative upstream section.
  • Primary decision areas: Pluggable Context Engine, Dual Compression System, 1. Gateway Session Hygiene (85% threshold), 2. Agent ContextCompressor (50% threshold, configurable), Configuration. Read those source anchors before changing installs, credentials, automation, or runtime configuration.
  • Useful search signals for this doc: context, compression, caching, system, anthropic. These are derived from the public title and summary so the clone remains lightweight.

Decision table

Best fitDeveloper Guide documentation

Use when you need orientation for context compression and caching before applying exact upstream commands.

Verify firstPluggable Context Engine

Open the source anchor for Pluggable Context Engine and confirm platform-specific requirements before changing configuration.

Escalate to sourceCommands, credentials, runtime behavior

Use the upstream page as the authority for current syntax, release changes, and security-sensitive steps.

Verification checklist

  1. Use this page to orient yourself within Developer Guide, then open the linked source page for exact current syntax.
  2. Start with the source section for Pluggable Context Engine; do not rely on the local summary for commands or secrets.
  3. Check platform, install method, provider, and credential assumptions before changing a real environment.
  4. Review the source section for Dual Compression System if the page involves setup, automation, messaging, or runtime behavior.
  5. After applying anything from the source, run the smallest relevant smoke test before widening scope.

Risk notes

Automation

Start with a bounded dry run, logs, and a manual stop path before enabling recurring or unattended execution.

External messages

Keep human review before sending public posts, customer messages, trading instructions, or team notifications.

Model spend

Set provider, token, and retry limits before scaling the workflow beyond a single test run.

Runtime access

Check filesystem, shell, browser, repository, and server permissions before granting the agent write access.

What this page covers

  • Core concept and where it fits in the Hermes Agent system.
  • Setup or operating context implied by the upstream page summary.
  • The source page link for full current details and updates.

Source mirror note

This page is generated from the public Hermes Bible index so the clone has the same route coverage and search surface. It stores the public title, category, summary, and source link locally; use the source page for full upstream text and updates.

Open source page