找到能用的 MCP 和 agent skills。

在 X(Twitter)上做舆情搜索与发帖的操作手册——x-manager 插件的通道选择(Grok 订阅 OAuth 与 xAI API key 自动降级)、自然语言代搜技巧、成本意识、结果边界、以及发帖前的确认纪律。需要搜 X 内容、判断该走插件还是浏览器、或要以用户身份发帖时读它。

Use when the user wants to see the UI or behavior an agent built, instead of launching the app and clicking through it themselves. Trigger on requests like "show me what you built", "demo this", "walk me through the UI", "screenshots of the change", "record the interaction", or /walkthrough. Captures stills and short clips of web, iOS, macOS, and Android front ends from the local build, in the background where the platform allows, and assembles them into a local HTML walkthrough with a one-command re-run script. Runs only on request, because captures are slow. Does not explain code: a request to understand how a change works, rather than what it does on screen, is not a walkthrough.

../../../engineering/minimalist/技能.md。

Triage any Telemetry 2.0 behavioral issue on RDK devices by correlating device log bundles with source code. Covers hangs, under-reporting, over-reporting, duplicate reports, CPU/memory spikes, scheduler anomalies, rbus problems, and test gap analysis. The user states the issue; this skill guides systematic root-cause analysis regardless of issue type.

Audit and analyze Snowflake warehouse, serverless, Adaptive, storage, transfer, and AI cost evidence with a typed ledger that prevents double counting and exposes freshness, attribution, control, and invoice boundaries. Use when a Snowflake bill increased, a team needs chargeback/showback evidence, credits appear unexplained, or an operator asks which usage merits investigation. Trigger with "Snowflake bill increased", "find idle Snowflake credits", "Snowflake cost attribution", or "untagged Snowflake spend". Do not use to mutate cost controls or claim savings.

Browser automation and web interaction using the opera-browser-cli tool. Use for navigating pages, clicking elements, filling forms, taking screenshots, inspecting console/network, running performance audits, and Opera AI features (chat available on any Opera browser; invoke-do, make, research require Opera Neon). When a browser is already running without automation enabled, this tool can restart it with a debug port (takeover) or use a separate profile — ask the user which they prefer.

Performs logistic regression for clinical trial outcomes (binary, ordinal, multinomial) with marginal-vs-conditional estimand reporting per FDA 2023 covariate adjustment guidance, g-computation/standardisation for marginal effects, modified Poisson for RR, Brant test for proportional odds, Firth penalty for separation, and Hauck-Donner detection. Use when modeling binary or ordinal endpoints in confirmatory or exploratory clinical trials.

Apply when writing tests, reviewing tests, deciding what to test, choosing test doubles, structuring test suites, or testing legacy code. Trigger on mentions of TDD, mocking, test coverage, test smells, integration tests, characterization tests, or getting code under test. Language-specific frameworks and idioms are in references/.