Codex long session, context budget, byte cap, command output protection, needle map, pre-compact raw data, living handoff, retrieval policy, repo workflow governance, PR gate, truncation recovery, large logs, large repo, stale memory, hallucination risk, token budget, oversized /goal, prompt compression, instruction placement. Use to govern evidence retrieval, output limits, handoff, and PR-stage control in long Codex repo/log/data sessions. Do not use for simple one-off tasks, tiny single-file edits, or tasks with no large repo/log/context/token risk.
Use this for repeatable market, competitor, company, product, or industry research tasks that require structured findings, source checking, and reusable research standards.
Use this when creating, improving, auditing, running retrospectives for, or updating Codex Skills. This Skill turns a repeatable task into a compact reusable SKILL.md and keeps Skill changes based only on verified lessons.