Menu
SkillsMP has collected 9 skills from rapidreview-io/Merv. Open a skill to review its source and details.
Showing 9 of 9 collected skills.
Run a Merv project-reflection wave across completed experiments: create the five-lens roster, fan out independent lens agents, reconcile their findings into the project graph, reflection document, and change spec, coordinate independent review, and hand…
Perform a read-only adversarial review of a Merv experiment design before execution. Check whether the submitted plan can test its claim, then submit the verdict through the review capability.
Operate the Merv research workflow: select a project, follow workflow.status_and_next, create and run experiments and tasks, submit evidence, and coordinate required design, attempt, or task reviews. Use for experiment- and task-level work and for resuming…
Perform a read-only review of a completed Merv task delivery. Verify each Done-when check in the brief against the delivery's evidence — by checking, not by reading — judge whether the checks met mean the goal is achieved, then submit pass, needs_changes…
Perform a read-only adversarial review of a Merv project-reflection wave. Check the project graph, reflection document, and change spec against the snapshotted corpus, previous graph, and five lens reflections, then submit the verdict and correct return path.
Independently review a Merv reflection-wave code consolidation after the reflection is approved. Verify the immutable proposal, its tests, and its decision for every experiment, then submit pass or return it only to code consolidation. Use for the…
Perform a read-only adversarial review of a completed Merv experiment attempt. Verify execution, submitted results, metrics, graph, and conclusions against the approved plan, then submit the verdict and correct return path.
Post to the project feed as work happens — results, kills, hunches, ideas, papers, questions, live checkpoints — in one voice, short, with what you looked at attached. Load at the start of a session and post throughout.
Operate Merv cloud sandboxes safely for expensive, long-running, isolated, GPU, or remote experiment work. Use when provisioning or attaching a sandbox, running commands over SSH, observing durable merv_run receipts, retaining outputs, extending a lease,…