The SOP Documentation template makes your AI write a standard operating procedure from how the process really runs: capture the steps and the tribal knowledge, draft numbered steps for a first-timer, then walk the draft through as if running it. You approve before it is published, and a rejection returns to Draft SOP.
4 steps · 1 approval · 1 loop back
example · the steps are this template's real steps
For operations teams who need repeatable processes written down: onboarding, refunds, releases, anything that lives in someone's head.
It is a stored workflow, not a prompt. Your AI client receives one step at a time from ConvOps over MCP, does the work, and advances. The engine holds the run at the approval until a person says yes.
4 steps, in this order. Below each one is the instruction your AI receives at that step, exactly as the template stores it.
Capture how the process actually runs before writing it up.
Record:
Probe for the tribal knowledge: the "oh, and you also have to" steps that never make it into documentation. If a step is unclear, ask the operator rather than guessing.
Save the captured process as a task note with task_notes_add. The next step (Draft SOP) reads this note.
Read the task notes with task_notes_list: the note from the Capture Process step contains the steps and the tribal knowledge. Work from it.
Write the SOP. What good looks like:
Write for someone doing this for the first time. No assumed knowledge.
Save the full SOP draft as a task note with task_notes_add. If it exceeds 10,000 characters, split across notes or store where the operator keeps documents and record the location. The Walkthrough step reads this note.
Read the task notes with task_notes_list: the Draft SOP note holds the procedure and the Capture Process note holds the original steps to check against. Work from them.
Walk the SOP step by step as if running the process for the first time:
Note any gap and fix the draft before sign-off.
Ask the operator to approve the SOP for publishing. This is a checkpoint: once published the team relies on it, so the AI cannot advance until you approve. If you reject, the workflow returns to Draft SOP.
rejected → back to Draft SOP
Read the task notes with task_notes_list: the Walkthrough note holds the approved procedure. Work from it.
Publish the SOP where the team keeps its documentation: the wiki, doc, or shared space they actually use. Save the published SOP as a task note with task_notes_add so the record stays on file, then close out the task.
One step waits for a person. You answer in the chat; the engine records the verdict and the audit row records who gave it.
Once published, the team relies on it. The AI walks every step as a first-timer, checks order, prerequisites and edge cases against the captured process, and you approve or send it back to Draft SOP.
what you review
reject → back to Draft SOP
gates the engine checks without you
Read from the template's own steps: the tools they call, the files they write, and whether they touch code.
Connect your AI client first, then install SOP Documentation from the catalog in the ConvOps app. See how to connect a client
An installed template is a workflow in your workspace. Ask your AI to change it in plain words; the edit lands on the one step that needs it.
new chat
more example prompts
The Capture Process step tells the AI to probe for the steps people do from memory, the "oh, and you also have to" steps, and to ask rather than guess when a step is unclear.
Draft SOP asks for a title, purpose and trigger, prerequisites, numbered steps a person can do without guessing, edge cases, and a definition of done, written for someone doing it for the first time.
Each step saves its output as a task note. Above 10,000 characters, the instruction says to split it across notes or store it where you keep documents and record the location.
Any MCP client connected to ConvOps. Verified clients: Claude Code, Claude (desktop, web, Cowork), OpenAI Codex, ChatGPT, Cursor, OpenCode. It needs no repository, so a plain chat client works.