'debug' 검색 결과

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.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

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

Build and debug sglang tests comparing token probabilities during prompt processing and step-by-step generation. Separate numerical differences from reused-state bugs, choose test thresholds, and locate the first operation that diverges.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Adding probability consistency tests for a model
• Choosing or justifying a kl_div threshold
• Separating numerical differences from cache state bugs

Configure how the browser starts, including window visibility, proxy settings, extra launch options, and a custom browser executable. It also explains opening a local debugging port so other local tools can connect.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Show the browser window during automation
• Start a browser with a proxy or incognito mode
• Choose a specific installed browser

Connect an agent to an existing Chrome browser, find local CDP ports used for remote browser control, or start a browser on a specified port for sharing with other agents and tools.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Connecting to Chrome with existing logins and open tabs
• Finding available local browser debugging ports
• Starting a browser on a fixed debugging port

Identify and unpack malware compressed with UPX to prepare executable files for static analysis, which examines code without running it. The workflow covers altered header repair, debugger-based recovery, and validation before analysis in Ghidra or IDA.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Identifying suspected UPX-packed samples
• Investigating failed UPX decompression
• Recovering binaries with a debugger

Compose compact, structured prompts for coding, debugging, review, and research with Codex or GPT-5.4. This internal skill defines task scope, completion criteria, output format, verification steps, and evidence requirements using consistent XML tags.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Define implementation scope and completion criteria
• Include verification steps in debugging requests
• Specify evidence requirements for code reviews

Keep project decisions, debugging notes, and session discoveries in a lasting Markdown wiki. Organize pages with tags and categories, connect related notes with links, and retrieve information through keyword searches.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Recording architecture decisions and their context
• Finding notes from earlier debugging sessions
• Saving discoveries at the end of a session

Use TensorBoard to visualize training loss and accuracy, compare experiments, and inspect model structure. Includes PyTorch and TensorFlow logging examples for examining weight distributions, predictions, and performance bottlenecks during machine learning training.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Comparing loss and accuracy across training runs
• Inspecting weight and gradient distributions
• Exploring embeddings in two or three dimensions

Use LangSmith to record LLM application inputs, outputs, and timing, investigate failures, evaluate answers against test datasets, and monitor production usage. It also guides automated quality checks and analysis of token usage and costs.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Investigating failures in LLM applications
• Evaluating answers against test datasets
• Monitoring production latency and costs

Get guidance on adapting LLMs with Unsloth, including implementation, API usage, and debugging. Covers LoRA and QLoRA, methods for adjusting models with fewer trainable parameters, with a focus on training speed and memory use.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Learning how to fine-tune models with Unsloth
• Exploring LoRA and QLoRA for efficient training
• Implementing features using Unsloth APIs

Configure Claude Code and improve how you assign coding tasks. This guide provides a CLAUDE.md template, practical prompting examples, debugging approaches using logs, and a workflow for making and checking small changes.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Set up project guidelines in CLAUDE.md
• Write clearer requests for coding tasks
• Investigate unexpected agent behavior using logs

Build custom cognee processing by chaining Python tasks, defining data types, and saving results to graph and vector databases. Use it to enrich existing data, inspect run status, and troubleshoot how tasks pass data.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Building custom extraction and processing steps
• Saving custom data types to graph and vector databases
• Adding information to an existing graph

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.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

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

Investigate failures in an existing Zoom integration by checking authentication, requests, SDK setup, media behavior, and MCP connections. The skill identifies the likely failing component and returns prioritized hypotheses, a short fix plan, and verification steps.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Investigate Zoom authentication and app configuration errors
• Check failed webhooks and event verification
• Diagnose SDK meeting joins and media session issues

Turn a Zoom integration idea into a recommended architecture and implementation path. The skill identifies the task, selects the appropriate Zoom technology, and routes requests to focused skills for authentication, meetings, bots, AI workflows, or debugging.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Choosing Zoom technology for a new integration
• Planning how to embed a Zoom meeting
• Finding the implementation path for a meeting bot

Investigate NanoClaw agent failures by tracing messages through logs and session databases. Check container startup, credential access, mounted folders, and service instances to locate where processing or delivery stops and choose a repair.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Tracing why an agent stopped replying
• Investigating container startup and authentication failures
• Checking mounted folders and stalled sessions

Set up, debug, and extend PicoClaw using its own commands and configuration structure. Get guidance on model authentication, external tool connections, scheduled tasks, and how conversations are routed and stored.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Setting up models and authentication in PicoClaw
• Configuring and troubleshooting MCP server connections
• Setting up recurring tasks

Fix bugs reported in GitHub issues through a reproduce-first process. The workflow guides code changes, build checks, verification, regression tests that check for recurring failures, and review before considering the fix complete.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Fixing a bug reported in a GitHub issue
• Reproducing a user-reported problem before editing code
• Verifying a fix and adding regression tests

Diagnose frozen or slow Qwen Code sessions on the current machine by checking CPU and memory usage, process states, child processes, network connections, and debug logs. Receive a diagnostic report with suggested next steps, without stopping processes.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Investigating an unresponsive Qwen Code session
• Diagnosing a specific process by PID
• Checking high CPU or memory usage

Explicitly save debugging discoveries, project conventions, and working preferences to Claude's persistent project memory. The skill checks for duplicate notes, keeps entries concise, and suggests moving enforceable rules into CLAUDE.md when appropriate.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Preserving hard-won debugging discoveries
• Recording undocumented project conventions
• Remembering tool-specific commands and pitfalls

Diagnose failing or intermittent Playwright tests by reproducing errors and examining execution traces. Classify the cause, apply a targeted fix, and run the test ten times to check stability, with recommendations to prevent recurrence.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Fixing a Playwright test that fails consistently
• Investigating tests that pass only sometimes
• Debugging failures limited to CI

Integrate Stripe payments and subscriptions into web applications, including trials, prorated plan changes, usage billing, customer portals, and invoices. Follow implementation patterns and local testing steps, with webhook signature checks and duplicate event handling to support reliable billing updates.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Add subscription checkout with a free trial
• Implement plan changes with a proration preview
• Add usage-based billing to a subscription product

Repair an entire broken feature by mapping its files and dependencies, checking configuration and logs, and confirming root causes. Apply fixes in sequence, then test the feature and the code that uses it.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Repairing a broken feature from start to finish
• Investigating imports and configuration together
• Finding root causes across multiple modules

Maintain cmux billing code with guidance on Stripe checkout, pricing, subscriptions, and access rights. It maps the relevant services and provides workflows for testing purchases, handling payment events, and rolling out billing changes.

한국어 요약은 준비 중입니다. 영어 요약을 표시하고 있습니다.

• Debugging checkout and subscription records
• Updating prices and plan switching
• Checking paid access and Cloud VM limits