AI Agent Hub
Back to plugins
🖥️

dsh-llms-forge

Client Updated 2026.08.25

Run the following command in DeepSeek Harness:

dsh plugin install zoahdev/dsh-llms-forge

Paste the following prompt into your AI chat to install this plugin:

Run dsh plugin install zoahdev/dsh-llms-forge inside DeepSeek Harness to install; the full source is at https://github.com/zoahdev/dsh-llms-forge

About this plugin

When an AI agent lands in a repository, the first file it looks for is usually llms.txt—a structured, machine-readable summary. Yet the vast majority of DeepSeek Harness plugins simply do not ship one: hand-written copies drift out of date, formats vary from repo to repo, and the absence of a file leaves agents to guess, slowing every downstream interaction.

dsh-llms-forge solves this with a single, deterministic pass over package.json and README.md. The generated file pulls its title from the package name, its summary from the description field (or the README first paragraph), and collects key facts such as install command, license, homepage, keywords, entry point, and CLI surface—without inventing any data. It is read-only by default, printing the proposed content to stdout; writing and overwriting are explicit opt-in flags. Exit codes distinguish clean runs, warnings, and usage or IO errors, making it straightforward to wire into a CI pipeline for consistency checks. There are zero runtime dependencies, so it leaves no trace after it finishes.

If you maintain a dsh plugin and want your repo to be agent-ready, or your team needs llms.txt validated on every merge, this plugin was built for that job. It does not replace human-authored documentation; it simply keeps the machine-facing entry file honest, reproducible, and auditable alongside your package metadata.

Use Cases

  • Generate an llms.txt for a plugin repo in one shot and write it to disk
  • Use exit codes in CI to verify llms.txt stays in sync with package metadata
  • Give AI agents a structured summary the moment they land in a repo instead of guessing

Best For

  • Developers maintaining dsh plugins who want their repo to be agent-ready
  • Teams that need llms.txt validated as part of a recurring CI pipeline
  • Engineering teams seeking reproducible, zero-dependency, auditable package metadata