digital-twins

Author a graded Pome task for a builder's agent and save it to their team catalog. Library-first — adapt an existing task before writing fresh; interview the builder about what would go wrong; draft criteria as code/model markers; validate + dry-run against the twins; then save_task. Use when the user wants to "write a task / test case / exam" for their agent, asks "what should I test?", or wants to turn a worry about their agent into a graded check.

Propose a builder's first Pome task by reading their already-registered agent — reads the pome.json manifest (twins) and the agent's prompt/code, proposes 2–3 candidate tasks grounded in what the agent actually does, runs a short interview to pick one, then hands off to pome-author-task. Use in a repo that already has a pome.json but no task yet, when the builder says "test my agent" or "what should I test?" and doesn't know where to start.

Intakes a Claude managed agent for testing on Pome — collects the full clone scope (agent YAML, environment config, attached memory stores, deployment kickoff events), registers it via the Pome control MCP's intake_clone_scope, and reports which of the agent's mcp_servers have twin coverage. Use when the user pastes a managed-agent YAML, says "test my agent with pome", or asks which of their MCP servers have twin coverage.

Entry point for testing an agent with Pome — routes to the right coach skill by context (managed-agent YAML → pome-intake, local repo / self-hosted REST agent → the local-examinee run path, plain task authoring → pome-author-task) and maps CLI-era commands to the hosted MCP tools. Use when the user says "test my agent with pome", "run pome", or "use pome". Supersedes the Gen-1 /pome-test skill.

Runs a verified Pome task against the builder's examinee and scores it from the live twin tape — run_task to mint the session, launch the examinee on its runtime (Managed Agents via ant, or REST), finalize_run the instant it idles while the tape is still live, then narrate get_report; re-runs only the failed tasks after a prompt fix and shows the delta. Use when the user's task passed seed verification and they want to run the exam, asks "run my tasks / how did my agent do?", or wants to re-test after a prompt fix.

Verifies a Pome task's seed is a fair exam before any run — verify_seed plus guard-aware triage of already-passing criteria, a state-diff review, an evaluate_criteria dry-run, and opt-in read-only probes on a live twin session. Use when the user has authored or adapted a task and wants to check the seed, asks "is my seed right / is this a fair exam?", or is about to run a task for the first time.