Hermes SkillForge logo Prescosoft private offline companion kit for self-improving agents
100% client-side localStorage persistence ZIP export Hermes · Claude Code · Codex · Cursor

AI Agent Skill Builder — Forge skills, memories & prompts for Hermes Agent.

Hermes SkillForge is a private browser workspace for shaping reusable AI-agent assets. Nothing is uploaded. Your work stays in your browser until you export it.

Skill Builder

Create a reusable Hermes-style skill document.

Live Markdown Preview

Memory Studio

Structure notes for persistent agent memory.

Structured Memory Output

Prompt Lab

Enhance prompts with looped reasoning, self-critique, tool discipline, and skill creation framing.

Optimized Prompt

Log Analyzer

Paste agent logs to parse tool calls, errors, retry loops, latency clues, and improvement ideas.

Analysis Report

Config Builder

Build a portable starter configuration for Hermes-style workflows.

Generated Config

Library

Saved skills, memories, prompts, log reports, and configs stored in this browser.

What Are AI Agent Skills?

An AI agent skill is a structured, reusable document that teaches an agent how to perform a specific task consistently. Unlike a one-off prompt that disappears after one conversation, a skill persists across sessions and defines:

Activation Triggers

When the agent should invoke this skill — user phrases, task types, or context patterns that signal the skill is relevant.

Procedures

Step-by-step operating instructions the agent follows, including tool selection, reasoning patterns, and decision criteria.

Examples

Sample inputs and expected outputs that anchor the agent's behavior and provide quality calibration.

Guardrails

Constraints, escalation rules, safety boundaries, and self-check procedures the agent must respect.

Skills vs. Prompts

A prompt is a single instruction for a single conversation. A skill is a reusable operating model — structured, versioned, and loaded by the agent whenever relevant. Skills turn ad-hoc prompt engineering into durable institutional knowledge your agent carries across every session.

Cross-Platform Compatibility

SkillForge exports standard Markdown skill documents compatible with Hermes Agent, Claude Code (.md skills), OpenAI Codex (AGENTS.md), Cursor (.cursorrules), and Gemini CLI. No proprietary lock-in — your skills are plain text you own.

Frequently Asked Questions

Common questions about AI agent skills, prompt engineering, and using Hermes SkillForge.

What are AI agent skills? ▾

AI agent skills are reusable, structured instruction documents that teach an agent how to perform specific tasks consistently. A skill typically includes a name, purpose description, activation triggers (when the agent should use it), step-by-step procedures, example inputs and outputs, and constraints or guardrails. Skills turn one-off prompts into durable, repeatable workflows that an agent can load and follow across sessions. They are used by Hermes Agent, Claude Code, OpenAI Codex, Cursor, and other agentic AI systems.

Is Hermes SkillForge compatible with Claude Code, Codex, or Cursor? ▾

Yes. Skills exported from SkillForge use standard Markdown format with trigger sections, procedure steps, and examples. This format is compatible with Hermes Agent, Claude Code skills (.md files), OpenAI Codex AGENTS.md, Cursor rules (.cursorrules), and Gemini CLI. The exported Markdown can be dropped directly into any of these tools' skill directories.

Does SkillForge work offline and keep my data private? ▾

Yes, all processing runs entirely in your browser using client-side JavaScript. Nothing is uploaded to any server. Your skills, memories, prompts, and configs persist in localStorage on your device and can be exported as Markdown, JSON, YAML, or ZIP files whenever you choose. No account, no tracking, no external AI API calls are made. This makes it ideal for prototyping proprietary agent workflows without exposing them to cloud services.

What can I build with the Prompt Lab? ▾

The Prompt Lab enhances raw prompts by injecting four optimization layers: full Hermes optimization (role definition, tool discipline, success criteria), looped reasoning chains (think → act → observe → decide), self-critique loops (draft → evaluate against criteria → improve), and skill creation framing (triggers, procedures, examples). Each enhancement can be applied independently or combined into a fully optimized prompt.

How does the Memory Studio help build better agents? ▾

The Memory Studio structures persistent agent context — facts the agent should recall across sessions. You define a memory title, descriptive tags, entity relations (people, tools, projects, constraints), and a raw note. SkillForge organizes this into a structured document that maps to how agents store and retrieve long-term knowledge, making context more organized and retrievable than free-form notes.

What does the Log Analyzer do? ▾

The Log Analyzer parses AI agent execution logs to identify tool calls, error patterns, retry loops, latency bottlenecks, and improvement opportunities. Paste any agent log output and it generates a structured report highlighting what went well, what failed, and where the agent's behavior could be refined — useful for debugging workflows and converting successful patterns into reusable skills.