DATA block to be filled by the build script (all ready skills, full skill.md content).
Everything below will light up once the data lands.
Use these guides to save time on work that helps your business grow. Pick one job, give AI the right facts, and check what it makes. Start with one draft you can review.
New here? See how to start one job. The ZIP includes START-HERE.md.
— SOPs. Their article mappings resolve to — hubs: — catalog-ready and — in progress. A Definitive marker also requires review of the actual page.
How we built this →Nothing in — SOPs matches that combination of search and status filter.
A skill is a written recipe; a pack groups recipes. A plugin is a package an app can load. An agent is the AI worker that uses them. This ZIP is a set of guides. Use the install guide for a supported app package.
“Reported complete” is the contributor’s saved label. It does not prove that we checked every part of the guide. It also does not mean a skill is installed, an account is connected, a job has run, or a business result was reached. Each needs its own evidence. Missing evidence stays unknown.
Every real run gets a meta article: a record of the work and its proof. Run records link to the recipe so we can improve it. Rechecking or editing the same run does not add a new run.
The Content Factory turns real stories into content that can help your business. Produce: gather the source. Process: shape it. Post: publish it. Promote: share what works. Setup and support tasks help these stages. Click a stage to find its tasks; read each guide for the actual inputs and next step.
Use each real run to check the recipe. Record what happened, including failed or blocked work, then review any lesson before changing the shared instructions. Writing the run record is required; public release needs its own authority.
Do → Check the result → Write the run record → Link the task → Review the lesson → ↺
One maintained recipe explains a repeatable task. Each actual execution gets a meta article: a written account of its steps, evidence, result, and next handoff. Keep the same execution ID across revisions and retries. A written draft is a valid record while public release is pending.
If a run reveals a useful lesson, propose a change to the recipe and review the evidence. The next execution tests an accepted change. A failed run, an unresolved gap, or no justified recipe change can be the honest result.