{"wiki":{"id":1636,"slug":"iv-s1","entity_type":"droid","entity_id":"iv-s1","title":"Initiative Validator","prose_md":"## About\n\nInitiative Validator is a reviewer droid owned by [[agt_011]]. Its job is to make sure every new initiative that wants to become active is grounded in today's Carol reality, not in stale assumptions from an older parent chain. Before an initiative can move forward, this droid checks each requirement, success criterion, and plan step against the current state of Carol’s systems — including [[sst]], [[carol-requirements]], [[carol-design]], [[build-cookbook]], and [[tests]] — and gives a verdict: keep, refine, or strike.\n\nIt works by loading the freshly authored initiative and a present-day snapshot of Carol’s state, then asking Claude (Haiku) to judge each item with a one-line comment. Refined items get a concrete replacement suggestion. Anything flagged is sent back to the Author for up to two refinement rounds; if items are still flagged after that, the droid logs a decision row noting that operator review is needed. Importantly, the Validator never edits the initiative itself — it only produces verdicts, and the Author remains the only writer. If the Validator fails, the Author’s draft is not blocked; a warning is simply logged.\n\nAlthough its exact name, \"Initiative Validator\", is unique, it is part of the larger reviewer droid family. That family is standard and reusable, with 31 droids run by multiple owners including [[agt_001]], [[agt_002]], [[agt_003]], [[agt_008]], [[agt_011]], [[agt_012]], [[agt_013]], [[agt_020]], [[agt_023]], [[agt_025]], [[agt_027]], [[agt_028]], and [[agt_031]]. So this droid is one specific instance of a shared pattern, not a one-off. It is not pure software: it calls Claude’s Haiku model to produce the keep/refine/strike verdicts, with the surrounding orchestration handled by the droid itself.","namesake_json":"{\"engine\": \"claude\", \"model\": \"haiku\", \"claude_purpose\": \"verdicting each initiative item as keep, refine, or strike\", \"family\": \"reviewer\", \"family_size\": 31, \"is_unique\": true}","profile_pic_path":"","source_hash":"9424b881c4f12fe1ae57a0b3d5b8ed77682afce1e8485b7970e84421f7307bef","status":"active","last_generated_at":"2026-08-05 00:37:17","created_at":"2026-07-29 03:33:16","updated_at":"2026-08-05 00:37:17"},"facts":{"id":"iv-s1","name":"Initiative Validator","machine_name":"IV-S1","owner":"agt_011","function":"Pre-flight verdict on every requirement, criterion, and plan step of a freshly authored initiative — keep / refine / strike — against present-day Carol state (CAROL-INI-509).","process_type":"triggered","schedule":"On demand (after Initiative Author)","process_name":"initiative_validator","avatar_color":"#a855f7","created_for":"Catching stale scope in freshly-authored initiatives before active work begins. INI-300-20's manager-droids test suite was the canonical regression — the Author drafted against a test landscape that no longer existed.","purpose":"Elrond must guarantee that every initiative entering active status is grounded in present-day Carol reality, not in stale parent-chain context. IV-S1 reads the initiative's requirements + criteria + plan steps, loads a snapshot of the live regression runner, current designs, cookbook, registry, requirements, and SST, and asks Haiku to verdict each item as keep / refine / strike. Flagged items are fed back to the Author for up to two refinement rounds.","duties":"- Loads the freshly-authored initiative (title, description, requirements, success criteria, plan steps).\n- Loads a present-day Carol state snapshot via shared.initiative_context.snapshot_current_state.\n- Calls Haiku to verdict each item: keep, refine, or strike, with a one-line comment and (for refine) a concrete replacement.\n- Returns flagged items to the orchestrator for Author refinement.\n- Round-trip is capped at two rounds; anything still flagged after the cap is logged as a decision row noting that operator review is needed.\n- Never edits the initiative itself — only produces verdicts. The Author is the only writer.","constraints":"- Only verdicts items already present on the initiative; never authors new items.\n- Cannot override the Author — only flags items for refinement.\n- Round-trip cap is two rounds; after that, the initiative is left for operator review with a clear decision row.\n- Validator failure (Claude call or JSON parse) does not block the initiative — the Author's draft remains and a warning is logged.","status":"running","gender":"male","archetype":"reviewer","building_block":"file","service_override":null,"enabled":1,"task_key":"initiatives.review_verdict","model_free":0},"page":{"type":"droid","page_class":"main","class_label":"Main page","kind_label":"Droid","kind_gloss":"","listed":true}}