orchestrator — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited orchestrator (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
降低学术文本 AI 模板痕迹,同时保持术语、引用、数字、公式、论断方向和章节功能不漂移。
核心原则:
先锚定事实与文意
-> 再全量识别 AI 模式
-> 再绑定真人写作实践卡
-> 再改写
-> 最后复扫残留先参考 ../../references/workflow-profile.md 获取 plugin interview、硬门禁、handoff artifact 和 verification evidence。该 profile 只提供高层路线参考,不锁死调用顺序;如果用户已经明确点名单个子 skill,仍直接走对应子 skill。
../../_shared/references/deweight-workflow-contract.md../../_shared/references/pattern-catalog.md../../_shared/references/semantic-anchor.md../../_shared/references/semantic-anchor-schema.md../../_shared/references/pattern-coverage-matrix-schema.md../../_shared/references/rewrite-prescription-rubric.md../../_shared/references/residual-verification-gates.md../../_shared/references/real-example-index.md../../_shared/references/anchor-db-schema.md../../_shared/references/anchor-curation-review-gate.md../../_shared/references/scoring.md../../_shared/references/safety-guardrails.md../../_shared/references/long-document-coverage-protocol.md../../_shared/references/human-writing-practice-cards/../../_shared/assets/academic_deweight_anchors.sqlite../../_shared/scripts/check_semantic_anchor.py, ../../_shared/scripts/check_pattern_matrix.py, ../../_shared/scripts/check_rewrite_prescription.py, ../../_shared/scripts/check_residual_report.pymode = aggressive | balanced | safe,默认 aggressive。target_tone = strict_academic | readable_academic,默认 readable_academic。segment_length,默认 450 中文字符或相近 token。report = rewrite_only | diagnosis | full_trace,默认 rewrite_only。deweight-intake-anchor.deweight-coverage-scan.deweight-human-anchor-bind for every suspect paragraph.deweight-anchor-rewrite.deweight-verify-gate.No step may be skipped unless the input is explicitly marked format-exempt and the reason is recorded.
The artifact chain must match deweight-workflow-contract.md. If a complete real example is requested and real-example-index.md says the workflow example is unavailable, report missing-real-workflow-example instead of inventing a case.
deweight-coverage-scan must use subagent-isolated scanning.blocked: needs_subagent_runtime
reason: long-document coverage scan requires isolated Local Scanner / Section Diagnostician / Global Aggregator / Anti-Miss Reviewer rolesDo not silently downgrade to single-agent long-document scanning.
Default output is rewritten text only after all gates pass. If report=diagnosis or full_trace, include:
Integrity Snapshot summary.Semantic Anchor summary.Pattern Coverage Matrix.Human Anchor Binding Table.deweight-verify-gate.When auditing or extending this suite, answer per child skill:
SKILL.md 保留:
references/ 新增或整理:
scripts/ 新增或整理:
assets/ 新增或不适用:
本轮能补:
仍缺 / blocked / needs-user:
判断依据来自哪些现有文件:suspect paragraph without human-writing practice card binding.anchor_review_decisions.decision=keep.uncertain mother skeleton that was not rescanned or explicitly left blocked.~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.