What problem does it solve? Keeping a personal task list inside an AI agent's memory is error-prone: tasks get duplicated, ambiguous references mutate the wrong item, and concurrent edits silently overwrite each other. This Skill manages the full task lifecycle (add, complete, defer, remove, review) against a single brain page with stable task IDs and fail-closed ambiguity handling. ## Core Features & Use Cases - Deterministic action routing: Maps user phrasing ("add", "done with X", "push X to next week") to exactly one lifecycle action before touching state. - Stable task IDs: Every task gets a t-YYYYMMDD-NN identifier so later actions target it unambiguously; fuzzy matches with multiple candidates return an ambiguous status instead of guessing. - Structured persistence: Tasks live in ops/tasks.md grouped by priority (P0-P3), with Deferred and Completed sections carrying dates and reasons; every mutation writes a timeline audit entry. - Use Case: Tell your agent "add task: renew the Acme contract, P1, due Friday", then later "complete the Acme renewal" — the Skill matches by ID, stamps the completion date, and returns a structured result other agents can chain on. ## Quick Start Ask the agent to add a task with a priority and due date, for example: add a P1 task to review the quarterly report due this Friday.
