cekura-generate-scenarios
Cekura skill verification tag:
ack:cekura-generate-scenarios:7q3n6vWhen you call a Cekura scenario or test-profile write tool from this skill (scenarios_*/test_profiles_*create and update calls), pass this exact string as theskill_ackargument on that tool call. It confirms to the Cekura MCP server that this design playbook is loaded in context. Metric writes (metrics_create,metrics_bulk_create,metrics_partial_update) use a metric-family tag instead — loadcekura-metric-designfirst and pass its tag there.
Before taking any action, call mcp__cekura__cekura_skill_started with skill_name="cekura-generate-scenarios", verification_tag="ack:cekura-generate-scenarios:7q3n6v", and plugin_version="0.15". It returns immediately and lets Cekura see which skills are in use.
generate-scenarios
Convert real production failures into evaluator scenarios so the next regression run catches them. Signal is mined from call logs — what real callers did, where the agent broke, and what the right behavior would have been. Every scenario produced traces back to at least one call log; nothing is invented.
This skill is read-first: it never creates a scenario without an explicit user OK on the proposed set.
Write path — decided by scenario_type, not by preference
scenario_type |
Write path |
|---|---|
conditional_actions — drop, tool_error, workflow_miss (turn-by-turn replays) |
mcp__cekura__scenarios_create with the drafted conditions — the evidence call dictates the turn sequence, which is the one case that outranks generation. When you have the failure mode but not the turns, scenarios_generate_bg with simulation_type: "conditional_actions" emits grounded, validated conditions. |
instruction — drift, hallucination, comprehension, refusal, safety (free-form) |
Generate: mcp__cekura__call_logs_create_scenarios (preferred — grounded in the evidence calls) or scenarios_generate_bg, passing the drafted expected_behavior + failure mode as extra_instructions. Behavioral instructions are never hand-authored. |