set-profilelisted

# Role Write `profile` field in state.json. Emit new profile summary. # Invocation /ultra-analyzer:set-profile <tier> [run-name] Where `tier` is one of: `small | medium | large | xl`. # Protocol ## Step 1: Parse arguments - First token of `$ARGUMENTS` = tier. Validate against enum {small, medium, large, xl}. Reject others with an error listing the valid options. - Second token (optional) = run-name. If absent, auto-detect single run per standard rules. ## Step 2: Profile effects Each tier declares downstream effects. Write them all into `state.profile` as a nested object so skills downstream can read without re-deriving: ```json { "tier": "large", "ultra_gate_tier": "--large", "worker_model": "sonnet", "worker_model_complexity_S": "haiku", "validator_model": "opus", "validator_model_complexity_S": "sonnet", "synthesizer_model": "opus", "topic_target_min": 45, "topic_target_max": 70, "redundancy_pair_rate_p1": 0.60, "suggested_parallel_terminals": "3-5" } ``` ### small ``` tier: small ultra_gate_tier: --small worker_model: haiku (all complexities) validator_model: sonnet validator_model_complexity_S: sonnet synthesizer_model: sonnet topic_target: 15-25 redundancy_pair_rate_p1: 0.20 suggested_parallel_terminals: 1-2 ``` ### medium ``` tier: medium ultra_gate_tier: --medium worker_model: sonnet (M/L), haiku (S) validator_model: opus validator_model_complexity_S: sonnet synthesizer_model: sonnet topic_target: 25-45 redundancy_pair_rate_p1: 0.40 su