Results for “search debug”

Battle-tested PyTorch training recipes for all domains — LLMs, vision, diffusion, medical imaging, protein/drug discovery, spatial omics, genomics. Covers training loops, optimizer selection (AdamW, Muon), LR scheduling, mixed precision, debugging, and systematic experimentation. Use when training or fine-tuning neural networks, debugging loss spikes or OOM, choosing architectures, or optimizing GPU throughput.

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

Run the AI Search Hub browser automation scripts for Yuanbao, LongCat, Doubao, Qwen, Gemini, Grok, and MiniMax. Use this skill when the user wants to ask one of those sites a prompt, auto-start or attach to a Chrome DevTools session on port 9222, seed an isolated debug browser profile from the user's local browser data, detect whether login is required, wait for the user to finish logging in if needed, and then continue automatically.

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

WeChat Mini Program development skill for building, debugging, previewing, testing, publishing, and optimizing mini program projects (小程序开发、调试、预览、发布). Covers project structure and config (`project.config.json`, `appid`, `miniprogramRoot`, `tabBar`, routing/navigation, icon assets), WeChat Developer Tools Nightly workflows (`wechatide` CLI, WeChat IDE Skills/MCP), `miniprogram-ci` preview/upload, console/network debugging, message push (消息推送) and customer-service auto-reply (客服消息), mini program SEO / search indexing (小程序搜索优化、页面收录、搜索推广、mpcrawler), and CloudBase integration (`wx.cloud`, 腾讯云开发, 云开发) when explicitly used. Use when users create, develop, modify, debug, preview, deploy, publish, or promote WeChat Mini Programs. NOT for Web frontend (use web-development), pure backend services (use cloudrun-development / cloud-functions), or UI-design-only tasks (use ui-design).

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

Answer a hard question by fanning it out to a PANEL of models running in parallel — each answering independently with web search and bash, none seeing the others' work — then having Opus 4.8 judge every response into a structured analysis (consensus, contradictions, partial coverage, unique insights, blind spots) and write a final answer grounded in it. The panel is two independent Opus 4.8 runs (slug opus4.8-4.8), Opus 4.8 + GPT-5.5 via codex (opus4.8-gpt5.5), Opus 4.8 + Gemini 3.1 Pro via agy (opus4.8-gemini3.1pro), or all three (opus4.8-gpt5.5-gemini3.1pro). Opus always judges and writes the final answer — the pipeline can't be reversed. Runs on local CLI subscriptions (no metered API), and saves a timestamped provenance .md per run. Use this whenever the user asks to "run it through Fusion", says /fusion, wants a multi-model / panel / ensemble answer, wants a question cross-checked across models, or wants a higher-confidence answer with consensus and blind spots surfaced — even if they don't say "fusion". General-purpose: any topic (research, law, strategy, technical, personal). Best for high-stakes research, design calls, and debugging where being confidently wrong is expensive.

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

Research and advise on the best way to approach a software project, including architecture, tech stack, implementation strategy, pricing/operating-cost tradeoffs, benchmark research, and comparisons with similar real-world projects. Use before building, mid-build, or after completion when the user asks for project strategy, optimal approach, research comparables, similar projects, stack selection, vendor/service choice, repo analysis, architecture critique, implementation feedback, or a prioritized improvement plan. Avoid for narrow single-bug debugging or isolated file edits unless the user asks for broader project direction.

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

Implement and execute approved research Experiment, Tuning, or Design Cards with reproducible logs, metrics, environment, and artifacts. Use for bounded coding, debugging, and runs; not for changing claims, priorities, budgets, or the Research Spine.

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

Use when the user needs to search or navigate code with Sourcegraph MCP tools. Provides disciplined search workflows for finding implementations, understanding systems, debugging issues, fixing bugs, and reviewing code.

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

Build applications and agents with Exa's API Platform: search, contents, answer, context, Agent API, monitors, websets, OpenAI-compatible endpoints, and exa-py / exa-js. Use when choosing Exa endpoints, writing Exa API calls, integrating semantic web search or research into products, or debugging Exa request shapes. Load references/ on demand for endpoint details.

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

Search and analyze Estuary task logs with jq filtering. Use when investigating task failures, debugging connector errors, analyzing behavior patterns, or streaming live logs. Use when user says "check logs", "show me logs", "task failing", "why is my capture failing", "why is my materialization failing", "error messages", "why did it fail", "search task logs", "shard failed", "follow logs", "increase log level", "debug logging", "is this log message normal", "what does this message mean", "connection timeout", or "i/o timeout".

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

Implement asynchronous work and changing data streams in Android and Kotlin Multiplatform apps using Kotlin Coroutines and Flow. Apply patterns for parallel loading, UI state updates, cancellation, error handling, and tests that check emitted values.

• Loading several data sources in parallel
• Managing screen state with StateFlow
• Handling search input delays and network retries

Diagnose REST and GraphQL API failures by checking connectivity, authentication, request formats, and returned data. Prepare reproducible commands, identify likely causes, and build regression tests that check whether a fix keeps working.

• Investigating API authentication failures
• Reproducing requests that work in Postman but fail in code
• Checking pagination and rate limits

Implement and debug large-model training with PyTorch FSDP, which splits training state across devices. Consult documentation and code patterns for sharding choices, mixed precision, and checkpoints when building or reviewing distributed training code.

• Implement distributed training for large models
• Compare FSDP wrapping and sharding patterns
• Review mixed-precision and checkpoint examples

Modify and troubleshoot services managed by s6, a process supervisor, inside the Hermes Docker image. Trace startup scripts, inspect profile gateways, add static services, and investigate restart failures or file ownership problems.

• Add a static service to the Hermes Docker image
• Investigate profile gateways that fail to start
• Debug gateway recovery after a container restart

Retrieve current documentation and code examples from Context7 for libraries, frameworks, and developer tools. Use the results to answer questions about API syntax, configuration, installation, version migration, and library-specific problems.

• Checking current API syntax
• Finding setup and configuration examples
• Investigating version migration issues

Find current documentation and code examples through the Context7 MCP server. Look up library setup, API usage, migration steps, and library-specific problems, using documentation for the requested version when available.

• Looking up documentation for a specific version
• Finding library setup and configuration steps
• Finding examples for code that uses external libraries

Search current library documentation and code examples with Context7 CLI, a command-line tool. Find API usage and configuration guidance, select available version-specific documentation, and prioritize examples in the programming language you use.

• Checking current API usage and configuration
• Finding documentation for a requested version
• Prioritizing code examples in a chosen language

Plan Claude Code agent teams by choosing team size, roles, and agent types with the tools each task needs. Preset compositions cover reviews, debugging, implementation, research, security audits, and migrations, alongside display settings for local or CI environments.

• Choose team size and roles for a development task
• Assign separate review areas to multiple agents
• Split bug investigation across competing hypotheses

Investigate SGLang runs that freeze across multiple GPUs. It uses stack traces, communication logs, GPU memory dumps, and comparisons between processes to locate state differences, trace their causes, and guide fixes.

• Locate where a multi-GPU inference run freezes
• Compare logs from distributed processes
• Trace the first difference in process state

Investigate production issues by searching Datadog logs, querying metrics, and following request traces across services. Compare error counts across time periods, inspect surrounding logs, and manage dashboards through a command-line tool.

• Investigating errors in production
• Comparing error counts across time periods
• Following request traces across services

Investigate bugs using a graph of code relationships. Find relevant functions, trace calls and execution paths, check recent changes, and assess what a proposed fix could affect before reading and changing the implementation.

• Narrowing down code related to a bug
• Tracing the execution path that triggers an issue
• Checking whether recent changes caused a bug

Search past coding sessions across Claude Code, Codex, and Cursor to recover earlier attempts, failed approaches, decisions, and findings. It filters history by topic and time, then summarizes technical details relevant to the question.

• Recovering approaches tried in earlier sessions
• Reviewing past debugging attempts and failures
• Summarizing recent work across coding agents

Build and maintain interfaces with shadcn/ui by finding, adding, and combining components. The skill uses project settings, component documentation, and change previews to guide styling, debugging, preset changes, and updates that preserve local edits.

• Building settings pages and dashboards from existing components
• Fixing form and dialog composition issues
• Updating components while preserving local edits

Turn plain-language requests into Cloud Logging queries for Google Cloud. Generate service-specific filters to find errors, investigate issues, or review audit logs using LQL, the language used to search these logs.

• Finding recent error logs
• Filtering logs for a specific Google Cloud service
• Investigating who changed or accessed a resource

Guides adapting OpenPI robot models to custom datasets and serving them to produce actions from observations. Covers JAX and PyTorch training, checkpoint conversion, and troubleshooting for normalization statistics and GPU memory.

• Fine-tune OpenPI models on custom robot datasets
• Convert JAX checkpoints to PyTorch
• Serve policies for ALOHA, DROID, or LIBERO