Results for “evaluation”

A meta-skill that understands task requirements, dynamically selects appropriate skills, tracks successful skill combinations using agent-memory-mcp, and prevents skill overuse for simple tasks.

The English summary is being prepared. Showing the original description.

Use after completing any non-trivial task. The agent self-rates its output on 5 axes — accuracy, completeness, clarity, actionability, conciseness — with concrete evidence per criterion. Produces a structured 1-5 scorecard with specific improvement suggestions.

The English summary is being prepared. Showing the original description.

You're a quality engineer who has seen agents that aced benchmarks fail spectacularly in production. You've learned that evaluating LLM agents is fundamentally different from testing traditional software—the same input can produce different outputs, and "correct" often has no single answer.

The English summary is being prepared. Showing the original description.

Standard Operating Procedures for read-only codebase exploration, symbol search, dependency mapping, code tracing, code review, architecture analysis, technical research, source evaluation, and evidence-based reporting.

The English summary is being prepared. Showing the original description.

Reviews the instructions text and skills of a declarative agent, with its stated capabilities, knowledge sources and audience, for prompt-injection exposure, data-leakage paths, over-broad permissions and missing refusals, and returns an attack-surface map, a findings table with severity, quoted evidence and attack path, one pasteable fix per finding, a probe pack the owner can run and a refusal-coverage table. Use when the user asks to "red-team this agent", "check these instructions for prompt injection", "could this agent leak data", "review the permissions on my agent" or "what should this agent refuse". Do not use for checking a SKILL.md against the format rules, use skill-file-reviewer instead; do not use for designing the launch test set, use agent-evaluation-plan instead. Drafts for human review; never approves, authorises or signs off.

The English summary is being prepared. Showing the original description.

Reviews one SKILL.md against the format rules and returns a findings table: strict YAML front matter with exactly name and description, name equal to the folder, description as the trigger, the eight-heading skeleton, capability-neutral wording, safety boundaries. Each finding carries severity, rule identifier, location, quoted evidence and replacement text; the verdict READY, REVISE or BLOCKED is an apparent state and the owner decides. Use when the user asks to "review this SKILL.md", "check my skill file before upload", "lint this skill description", "validate the front matter of this skill" or "score this skill folder". Do not use to write an agent's instructions field, use agent-instructions-drafter instead; for injection or leakage exposures use agent-instructions-red-team; for launch tests use agent-evaluation-plan. Drafts for human review; never approves, authorises or signs off.

The English summary is being prepared. Showing the original description.

Audits a live website for AI-engine discoverability (AEO/GEO). Crawls the site, runs 16 deterministic checks plus a 6-dimension content evaluation, and produces a scored report (A-F) with prioritized fixes. Use to get a baseline before improve-aeo-geo, or to measure progress after changes.

The English summary is being prepared. Showing the original description.

Simulate a Nature-style reviewer assessment from the referee perspective rather than an author rebuttal. Use when the user wants a pre-submission review, reviewer report, peer-review style critique, novelty/significance/technical soundness assessment, reviewer-style manuscript evaluation, 审稿人视角评估, 预审稿意见, or Nature reviewer report. Return 3 reviewer reports plus a cross-review synthesis, grounded only in the local Nature reviewer source basis. Also trigger on general pre-submission review requests during academic writing even without the word "Nature", such as getting a mock peer review for any journal, critiquing a draft as a reviewer would, assessing novelty/rigor before submission, and Chinese phrasings like 审稿人视角、模拟审稿、预审、帮我审一下论文、投稿前自审、审稿意见模拟、找论文问题.

The English summary is being prepared. Showing the original description.

Knowledge pack synthesized from JSP 376 Defence Acquisition Safety Policy V1.1 (April 2024, Ministry of Defence, Open Government Licence v3.0): acquisition safety requirements, safety artefacts, and their application across the standard acquisition lifecycle. Covers the policy frame and legislative baseline, SRO and User responsibilities, safety management and safety requirements, test and evaluation and the safety case, supervision and approvals decision support, and the seven lifecycle phases. SCOPE LIMITS: single-source (JSP 376 V1.1); MOD policy and guidance, not statute; builds on JSP 815 and cites JSP 655, neither reproduced here (use jsp-815 for the SMS); Chapter 4 (Non-Standard Acquisition) is unpublished in V1.1 and is not in this pack; no DEF-STAN text (Def Stan 00-056 is named, not quoted); no other JSP reproduced. Reconstructed reference notes, not MOD guidance, and no official status.

The English summary is being prepared. Showing the original description.

Plan and prepare an OpenAmer-specific fine-tune from the collected brain data: data curation, ChatML JSONL, training, evaluation. 'The organism becomes its own brain'.

The English summary is being prepared. Showing the original description.

Hijacking, Red-Team Evals, and Continuous Evaluation A technical reference on agent hijacking, red-t

The English summary is being prepared. Showing the original description.

Econometrics skill for Difference-in-Differences (DID) analysis. Activates when the user asks about: "difference in differences", "DID", "DiD", "diff-in-diff", "parallel trends", "treatment group", "control group", "pre-treatment", "post-treatment", "policy evaluation", "natural experiment", "staggered DID", "event study regression", "two-way fixed effects DID", "callaway santanna", "sun and abraham", "双重差分", "倍差法", "平行趋势", "处理组", "对照组", "政策评估", "事件研究", "交错DID", "渐进处理"

The English summary is being prepared. Showing the original description.

Knowledge base from NASA System Safety Handbook Volume 2 (NASA/SP-2014-612). Use for risk-informed safety cases (RISC), integrated safety analysis (ISA), adequate safety concepts, safety objectives and requirements, acquirer/provider roles, graded safety analysis, claims-evidence-argument structure, SSMP development, and RISC evaluation. Does not cover nuclear or chemical safety domains, occupational health regulations, or NASA-specific programmatic requirements beyond system safety.

The English summary is being prepared. Showing the original description.

Knowledge base from NASA-STD-8729.1A — NASA's R&M (Reliability and Maintainability) standard covering spaceflight and support systems (Revision A, 2017-06-13). Use for NASA's objectives-driven R&M framework — the R&M objectives hierarchy (one Top Objective decomposed into four sub-objectives, paired with tailorable strategies), how R&M requirements are established inside the SMA Plan required by NPR 7120.5, the SMA Technical Authority concurrence/independent-evaluation governance, milestone vs. readiness review gates, the reliability/maintainability/availability vocabulary (Ai vs. Ao, the failure causal chain, risk as a triplet), and the R&M Evidentiary Methods catalogue (FMEA/FMECA, FTA, RBDA, RCM, LORA, plus the space-environment and parts-pedigree analyses). Scoped to NASA spaceflight and support systems and to assurance planning: it is an objectives-and-strategies standard, so it deliberately does NOT prescribe step-by-step analysis procedures (those live in the referenced NASA Preferred Reliability Practices) and is thin on facility R&M, detailed math/derivations, and non-NASA or commercial life cycles.

The English summary is being prepared. Showing the original description.

Knowledge pack from 'Towards Digital Engineering: The Advent of Digital Systems Engineering' (Huang et al., Old Dominion University, arXiv:2002.11672, 2020). Use when reasoning about digital engineering, the DoD Digital Engineering Strategy and its five goals, the Authoritative Source of Truth, digitalization vs. digitization, digital augmentation, digitalized models, provenance, unique identification, and the four-level (vision/strategy/action/foundation) DSE research framework. Covers the conceptual/vision layer of the field. SCOPE LIMITS: this is a single ~28-page vision paper, not a methods handbook — it is thin on step-by-step procedures, tooling, case studies, and quantitative evaluation, and it predates 2020 so it does not cover later digital-engineering standards or recent AI/LLM advances. Use the sebok pack for established SE canon and MBSE depth.

The English summary is being prepared. Showing the original description.

Knowledge base from the DoD Test & Evaluation Enterprise Guidebook (August 2022 edition 8.02) with DOT&E-facing enterprise emphasis. Use for T&E enterprise roles (DOT&E vs USD(R&E)), DT&E/OT&E/LFT&E, cybersecurity T&E, MOSA/automation/digital enablers, suitability/reliability growth, and TEMP/Strategy/STE planning. Distinct routing from dod-te-guidebook (same guidebook family, pathway-deep pack). Guidance only—defer to DoDI 5000.89 and Title 10 for binding requirements.

The English summary is being prepared. Showing the original description.

Knowledge base from the FAA System Safety Handbook (Dec 30, 2000). Use for system safety as a specialty within systems engineering: the five-step safety risk management process (mandated by FAA Order 8040.4 and the AMS), the severity-by-likelihood risk matrix and acceptance bands, the Safety Order of Precedence, system-description models (5M, SHEL(L)), the hazard-analysis activities (PHL/PHA/RHA/SSHA/SHA/O&SHA/HHA), integrated system hazard analysis, the analysis techniques (FTA, FMEA/FMECA, Fault Hazard, Common Cause, Sneak Circuit, Energy Trace), closed-loop hazard tracking (Safety Action Record), and the specialty domains the handbook covers — acquisition-lifecycle safety and contracting (OSA/CSA/SSPP/ISSPP), software safety (control-capability ranking, SHCM, DO-178B), test & evaluation safety, facilities safety, commercial launch safety (AST licensing, expected casualty, MPL/DAMP), system safety training, and Operational Risk Management (ORM) with human/organizational factors. Scope limits: this is a year-2000 FAA how-to; it predates the modern FAA/ICAO Safety Management System (SMS) and Just Culture regulatory frameworks, and many cited orders, advisory circulars, and launch-licensing rules have since been restructured. It is FAA/NAS-centric, leans on MIL-STD-882C, and does not define the SE life-cycle processes themselves (use nasa-npr-7123 / dau-se-guidebook) or reproduce third-party standard text (MIL-STD-882, DO-178B, ISO/IEC/IEEE).

The English summary is being prepared. Showing the original description.

Structured problem solving in 6 steps: Problem definition, goals, brainstorming, evaluation, implementation, and review.

The English summary is being prepared. Showing the original description.

Structured decision making: Pro/Con matrix, weighted scoring, decision tree, scenario analysis, and Eisenhower matrix.

The English summary is being prepared. Showing the original description.

当用户在评估"这个诱因好不好""这个场景绑定有没有效""咖啡和热巧克力哪个是更好的诱因"时激活此skill。 典型触发信号：用户提到"评估诱因""这个场景绑定行不行""选什么作为品牌诱因""诱因效果好不好""哪个刺激物更适合做激活线索"等。 不适用于：还在找诱因的阶段（那是trigger-habitat的工作）、评估内容的趣味性（那是social-currency的工作）、评估情绪感染力（那是情绪skill的工作）。 有效诱因评估矩阵：频率（诱因多久出现一次）、强度（诱因是否专一指向产品）、情境（诱因激活时用户能否采取行动）三维量化评估。频率比强度更重要。

The English summary is being prepared. Showing the original description.

Google Sheets skill for reading and writing spreadsheets via the gws CLI. Covers cell reads, row appends, batch updates, sheet management, and formula evaluation.

The English summary is being prepared. Showing the original description.

Architectural decision-making framework. Requirements analysis, trade-off evaluation, ADR documentation. Use when making architecture decisions or analyzing system design.

The English summary is being prepared. Showing the original description.

Overview of six AI engineering workflows: prompt evaluation (8-dimension scoring), context budget planning, RAG pipeline design, agent security audit (65-point checklist), eval harness building, and p

The English summary is being prepared. Showing the original description.

Use when comparing prompt changes against a fixed evaluation set: record actual model settings, required checks, missing results, judge limitations and case-level regression evidence.

The English summary is being prepared. Showing the original description.