Orchestrate research entry lifecycle in ./research/ — create, batch-import, refresh stale entries, and validate structure. Use when asked to add a tool, research a URL, document a library, refresh research, validate entries, or given any tool or library URL. Supports --batch (parallel multi-URL), --rerun (refresh one or all entries), and --validate (structural check with auto-fix of error-severity issues).

Audit Snapchat Ads measurement, Snap Pixel and Conversions API, mobile and app campaigns, creative, AR and catalog formats, audiences, budgets, brand safety, and reporting. Use for Snapchat Ads, Snap Ads, Snap Pixel, Snapchat Conversions API, AR Lens ads, app-install campaigns, or Snapchat dynamic product ads.

Monitor paid-ad account pacing, delivery, performance, creative fatigue, tracking, policy, and data quality across supported platforms. Use for daily or weekly checks, anomaly review, budget pacing, post-launch verification, or campaign monitoring.

Audit LinkedIn Ads measurement, Insight Tag and conversions, professional audiences, lead generation, ABM, creative, bidding, budgets, pacing, automation, and policy. Use for LinkedIn Ads, Campaign Manager, Insight Tag, Lead Gen Forms, Thought Leader Ads, ABM campaigns, or B2B paid media.

Transforms conversations and discussions into structured documentation pages in Notion. Captures insights, decisions, and knowledge from chat context, formats appropriately, and saves to wikis or databases with proper organization and linking for easy discovery.

Workflow 4: Submission rebuttal pipeline. Parses external reviews, enforces coverage and grounding, drafts a safe text-only rebuttal under venue limits, and manages follow-up rounds. Use when user says "rebuttal", "reply to reviewers", "ICML rebuttal", "OpenReview response", or wants to answer external reviews safely.

DEFAULT poster pipeline — build an academic conference poster (ICML/NeurIPS/ICLR/CVPR/...) as a single HTML/CSS file with measurement-driven hard gates, real paper figures, a two-hue design-token system, and print-ready PDF via headless Chromium. Use when the user says "做海报", "poster", "conference poster", "paper poster", or asks to design/redo a research poster.

Audit a live page for accessibility issues, locate each WCAG violation precisely, and return a selector-grounded fix worklist without editing.

Vercel Queues guidance — durable topics with at-least-once delivery, independent consumer groups, retries, delays, and idempotency keys via @vercel/queue (JS) or vercel-queue (Python). Use when deferring background work, buffering traffic, fanning out events, or choosing between Queues and Workflows.

Authentication integration guidance — Clerk (native Vercel Marketplace), Better Auth, Descope, and Auth0 setup for Next.js applications, plus Sign in with Vercel, Vercel Passport, and Vercel KMS. Covers server config, route handlers, proxy.ts auth patterns, sign-in/sign-up flows, and Marketplace provisioning. Use when implementing user authentication, sessions, protected routes, or protecting deployments.

Update the Claude Code Ultimate Guide's CLI release tracking and optionally bump the guide version. Use for 'mets à jour le suivi des releases Claude Code dans le guide', 'synchronise la dernière version CLI avec le landing', or the update-infos-release command. Not for Codex CLI releases, checking installed software updates, writing a generic changelog, or publishing without authorization.

Use when a change is non-trivial by DSH standards (behavior, architecture, cross-file contracts, process/tooling, testing strategy, or on-disk/wire/config formats), when choosing between technical alternatives, superseding a decision, or writing a postmortem. Records the why and rejected options in .agents/notes/ with script-enforced gates; skips purely mechanical edits (CRUD, styling, patches, tagging, formatting).

Use when the user mentions Foundation Models review, on-device AI audit, LanguageModelSession issues, or @Generable/guardrail problems — including missing availability checks, manual JSON parsing, and guardrail handling.

Use when creating Codex, Claude Code, Gemini CLI, Cursor, or AGENTS.md runtime adapters from one canonical agent core. Use whenever a generated repo needs multiple AI runtimes without duplicating instructions.

Use when packaging an Agentlas agent repo for public GitHub release, Codex plugin submission, Claude adapter distribution, or one-line terminal installation.

Explore unfamiliar APIs, libraries, or implementation behavior with minimal reproducible examples and documented findings. Use when implementation details are unclear and can be resolved by reading docs, inspecting code, and running focused experiments.

HashQL diagnostic writing patterns using hashql-diagnostics crate. Use when creating error messages, warnings, Labels, Messages, Severity levels, Patches, Suggestions, or improving diagnostic quality in HashQL code.

Amazon Redshift is NOT PostgreSQL; corrects LLM mistakes; covers SQL, metadata, connectivity/auth, storage, query performance, and operations. Applies ONLY to Redshift (provisioned or Serverless). Pushes back on: CREATE INDEX, string_agg, pg_catalog, text type, SERIAL, stl_query. Triggers on: Redshift SQL, Redshift CREATE TABLE, Redshift COPY/UNLOAD, slow or suddenly slow Redshift query, EXPLAIN plan, queue wait, stale statistics/ANALYZE, result cache, Redshift permission denied, Redshift disk full, storage growth, snapshot/backup cost, ghost rows/bloat, VACUUM, spill to disk, distribution skew, Redshift system views, QUALIFY, PIVOT, MERGE, materialized views, Redshift Data API (Python/boto3), Redshift workload management (WLM), concurrency scaling, Redshift resize, Spectrum, awsdatacatalog, connection failures (SSL/TLS, ODBC/JDBC, connection limit), IAM/SAML federation, IAM Identity Center, datashares. Does NOT apply to: S3 policies, Athena, Glue/Iceberg outside Redshift, Aurora, RDS, DynamoDB.

Console 前端交付范式：4 道门禁驱动的前端开发流程。Use when: 新增前端能力、settings section 迁移、新增页面、重构布局、或 F190/Console 级前端流程需要 Product/Design/Implementation/Verification gate。 Not for: 小样式点改、纯后端 API、纯计算逻辑、独立 Design System token 定义。 Output: 通过 Product / Design-System / Implementation / Verification gate 的前端代码与证据。

Design new Claude skills from structured idea specifications. Use when the skill auto-generation pipeline needs to produce a Claude CLI prompt that creates a complete skill directory (SKILL.md, references, scripts, tests) following repository conventions.