Verify
The reviewer runs the verification written in the phase spec and reports what passed and what failed. Reject it and the run returns to Implement. Issues are never patched inline.
reject → back to Implement
The Phase template makes your AI execute one chunk of a larger plan: confirm the phase spec is clear and its dependencies are met, implement inside that scope, then run the phase's own verification. You approve the result. A failure loops back to Implement and verification reruns from scratch.
4 steps · 1 approval · 1 loop back
example · the steps are this template's real steps
For teams running multi-phase plans, where each phase is one task an AI developer can pick up cold.
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.
Read this phase's specification from the parent plan: typically a file under the plan's docs directory describing scope, approach, and verification for this phase.
Confirm:
If anything is unclear or a dependency is unmet, raise it now before starting implementation.
Execute the work described in the phase specification. Keep changes focused on this phase's scope: do not scope-creep into adjacent work.
Run tests after each significant change to catch regressions early.
Run the verification defined in the phase specification. Report what you checked, what passed, and what failed.
If anything fails, reject to loop back to implement. Do not patch issues inline: the developer fixes, then verification re-runs from scratch.
Request operator approval before advancing.
rejected → back to Implement
Phase complete. The parent plan's execute-phases step will detect this and advance to the next phase automatically.
One step waits for a person. You answer in the chat; the engine records the verdict and the audit row records who gave it.
The reviewer runs the verification written in the phase spec and reports what passed and what failed. Reject it and the run returns to Implement. Issues are never patched inline.
reject → back to Implement
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 Phase 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
From its parent plan. The Ready step reads the phase file under the plan's docs directory (scope, approach, verification) and raises anything unclear or any unmet dependency before implementation starts.
The parent plan's Execute Phases step sees it. That step waits until every child phase is completed or cancelled before the plan moves on to its own verification.
It attaches to the phase task type, so it runs for any phase task. Its Ready step expects a spec from a parent plan, so it is designed to run under the Plan template.
Any MCP client connected to ConvOps. Verified clients: Claude Code, Claude (desktop, web, Cowork), OpenAI Codex, ChatGPT, Cursor, OpenCode. Because its steps change code, run it in a coding client with your repository checked out.