A bounded self-evolving career-search agent that combines Gödel-style policy evolution with a Field Onboarding Map for job discovery, company/product due diligence, application strategy, interview conversion, and durable memory.
This GitHub repository is the single source of truth. Google Drive is treated only as migration history and is no longer authoritative.
- Reach qualified interviews within the active search window.
- Prioritize companies with credible industry and product prospects.
- Keep overseas sales, multinational commercial roles, enterprise AI/data/cloud, robotics, industrial technology, and adjacent bridge roles open.
- Convert user feedback into small, tested policy improvements rather than an ever-growing prompt.
- Preserve evidence, rollback, and evaluator independence.
PROJECT.md— current project state and operating goal.SKILL.md— executable agent skill.docs/ARCHITECTURE.md— system map and components.docs/CONSTITUTION.md— protected invariants.docs/EVOLUTION_PROTOCOL.md— bounded Gödel evolution.memory/STATE.md— current state and interview funnel.memory/PREFERENCES_AND_COMPANY_LEDGER.md— stable preferences and company states.
Ordinary daily runs may update runtime memory, but must not edit the policy, evaluator, or constitution. Policy changes happen only through a dedicated evolution branch or pull request with baseline/candidate evaluation and rollback.