
from Liuyishou8
A personal career-defense skill to capture work evidence, create weekly reviews, and generate manager-facing packs and performance bullets from day-to-day activ
Liuyishou helps agents capture short, verifiable records of recent work (proofs), preserve tacit judgment, and convert entries into weekly reviews or reusable artifacts like manager updates, performance bullets, mobility packs, and exit packs. It prescribes a storage model (.liuyishou/), fast capture commands, and deterministic export patterns so outputs are reproducible and auditable.
Use this skill when the user wants to quickly log a task, incident, launch, or decision; produce a weekly defense review; assess AI-compression risk for recent work; or assemble performance/mobility/export packs for managers or job moves. Avoid fabricating achievements or including private reasoning in shareable exports.
Designed to work with assistants that can run local scripts and read/write files (Claude-style or LLMs with a filesystem tool integration). The skill expects deterministic script outputs and LLM-based rewriting for audience-facing copy.
Career-defense skill that helps workers capture work evidence, generate weekly reviews, and produce manager/performance packs. The Python script is well-structured with argparse subcommands for init/capture/weekly/moat/pack, uses only stdlib, and stores data locally in .liuyishou/. Script exits with code 2 when no subcommand is provided (expected under DRY_RUN). No security concerns — purely local file operations with no network calls or credential handling.
Clean, privacy-first design. No exfiltration risk. The SKILL.md explicitly forbids fabricating achievements and instructs separation of shareable vs private data. The script failed only because DRY_RUN provides no subcommand args, which is expected behavior.