Octonous Adds Markdown Agent Skills for Shared Guidance Across Chats and Automations

Octonous has added Agent Skills, a library for reusable instructions and context stored in Markdown files. The change gives users a way to maintain common guidance once, then load it into conversations or attach it to automations rather than rebuilding the same setup for each task.

A Skill is a named set of instructions or context that an agent can load when it is relevant. Octonous positions the files for material that spans work: writing rules, product information, research methods, operating procedures, review criteria and organizational context. The files use plain Markdown and follow the open Agent Skills format.

One library, two ways to use it

The product draws a line between an automation and a Skill. An automation decides when a task runs; a Skill supplies reusable guidance for how it runs. A scheduled newsletter draft every Thursday, for example, can use product-positioning context that is also available to other automations and to a live conversation.

Getting guidance into the library

1. Write a Skill manually by supplying its name, description and instructions.
2. Use Octonous’s built-in skill-builder, which asks clarifying questions about the workflow, intended outcome and constraints before producing a draft for review and approval.
3. Import a SKILL.md file by providing its GitHub URL or uploading the file directly.

The import option matters because it accepts a file-based format rather than requiring every team to recreate its guidance inside Octonous. The company says imported and newly created Skills can be kept in a personal or team-wide shared library.

Sharing changes the operating model

New Skills are private by default. A user can share one with a workspace, giving colleagues a common source of instructions and context; when that shared material changes, it can be updated in one place instead of maintained as separate copies by each user.

Skills are available to all Octonous users. They can be invoked in a conversation with a slash command, while attached Skills are used on each scheduled automation run. The practical choice for a team is therefore not whether a prompt can describe a task, but whether recurring instructions deserve a maintained, shared file of their own.