Gentlybot docs

Knowledge & skills

Knowledge tells Gentlybot what your team knows. Skills teach it a repeatable way to work. Together they make good results easier to reproduce.

Shared facts, context, and decisions

  • Project knowledge is the default context; organization-wide knowledge holds conventions shared across projects.
  • Create blank pages, runbooks, bug analyses, FAQs, and nested subpages.
  • Edit visually or as Markdown, add images, and link pages together.
  • Use backlinks and page history to understand what changed and restore an earlier version.
  • Agents working in a project can read and update that project’s pages directly.

Reusable instructions and capabilities

  • Search the skills library by team or discipline and preview the instructions before enabling anything.
  • Enable a skill so agents can discover it automatically, or run it explicitly with its slash command even while disabled.
  • Upload a workspace skill, or commit a SKILL.md alongside the repository it belongs to.
  • Store any credentials a skill needs in workspace-managed skill secrets instead of embedding them in instructions.

Is this information about our product or decisions?

Put it in Knowledge: architecture notes, terminology, support runbooks, release process, known failure modes, or product FAQs.

Is this a repeatable way of doing work?

Make it a Skill: a checklist, audit method, reporting workflow, integration procedure, or team-specific deliverable format.

Is it specific to one repository?

Keep the skill or knowledge close to that project. Use organization scope only for guidance that should apply across products.

Start small: one product glossary, one “how releases work” page, one support runbook, and one skill for a task your team repeats every week. Add more only when a real thread exposes missing context or a repeated process.

Keep facts in Knowledge and actions in Skills. That separation makes both easier to review, update, and reuse.