Repolode
Submit
Cranot

Cranot/super-hermes

Hermes Agent skills that have the agent write its own analytical prompts (prisms) tailored to each artifact, then report both findings and blind spots.

View on GitHub
Stars
498
Forks
31
Open issues
0
Last commit
3 months ago

What is it?

Super Hermes is a set of skills for Hermes Agent that makes the agent generate its own analytical instructions before tackling a complex task. Each generated instruction, called a prism, is tailored to the specific artifact, such as a source file, a business plan or a strategy document, and the agent then executes it. The output covers what was found along with what was not examined.

The project ships five slash-command skills: prism-scan for a single tailored analysis, prism-full for a multi-pass pipeline where later passes challenge earlier findings, prism-3way for a where/when/why analysis, prism-discover for mapping analysis domains, and prism-reflect for a constraint report on the analysis's blind spots. It also includes seven pre-built prisms that can be used as system prompts with other LLM tools.

The skills save constraint notes to a .prism-history.md file in the project, so later analyses in the same project can adjust their approach to cover previously missed angles. Installation is available through the Hermes skills hub or by cloning the repository and running an install script.

Key features

  • 01Generates a custom analytical prompt for each artifact before executing it.
  • 02Reports both findings and an explicit list of what the analysis did not examine.
  • 03Includes a multi-pass mode where later passes challenge earlier conclusions.
  • 04Provides seven pre-built prisms usable as system prompts with any LLM tool.
  • 05Stores constraint notes per project to guide future analyses.

Use cases

  • Reviewing a source file for structural trade-offs and concrete bugs with line references.
  • Stress-testing a business plan or strategy document from several angles at once.
  • Checking an analysis for blind spots before relying on it.
  • Running pre-built analytical prompts with Claude CLI or the Anthropic API.

Quick start

  1. 1

    Install the prism-scan skill via Hermes Agent

    hermes skills install Cranot/super-hermes/skills/prism-scan
  2. 2

    Run the skill on a file inside Hermes

    /prism-scan analyze examples/circuit_breaker.py

Things to know

  • Analyzes one file at a time; cross-file architectural analysis is not supported yet.
  • Output quality depends on the underlying model, and requires Hermes Agent to run the skills.
  • Growth mechanism is described as minimal in the README.

Topics

#ai-agent#prompt-engineering#code-review#llm#hermes-agent#static-analysis

FAQ

What is super-hermes?
Hermes Agent skills that have the agent write its own analytical prompts (prisms) tailored to each artifact, then report both findings and blind spots. Super Hermes is a set of skills for Hermes Agent that makes the agent generate its own analytical instructions before tackling a complex task. Each generated instruction, called a prism, is tailored to the specific artifact, such as a source file, a business plan or a strategy document, and the agent then executes it. The output covers what was found along with what was not examined.
Is super-hermes free to use? What license does it have?
super-hermes is open source under the MIT license. Check the license terms before using it in commercial products.
Is super-hermes still maintained?
Updated within 6 months. The last commit to super-hermes was on Jul 27, 2026.
What can I use super-hermes for?
Reviewing a source file for structural trade-offs and concrete bugs with line references. Stress-testing a business plan or strategy document from several angles at once. Checking an analysis for blind spots before relying on it. Running pre-built analytical prompts with Claude CLI or the Anthropic API.
How do I install super-hermes?
Install the prism-scan skill via Hermes Agent: hermes skills install Cranot/super-hermes/skills/prism-scan Run the skill on a file inside Hermes: /prism-scan analyze examples/circuit_breaker.py
Where can I get super-hermes?
The source code and installation instructions are on GitHub: https://github.com/Cranot/super-hermes

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects