The Art of Debugging
stas00/the-art-of-debugging
Condensed debugging method and tool recipes for Unix, Python and PyTorch programs: crashes, hangs, segfaults, wrong output, CUDA OOM, NaN values and slowness.
Measures and shrinks the ExecuTorch runtime binary by building a size test, analyzing it with bloaty and landing each reduction as its own pull request.
Install ExecuTorch Binary Size Reduction
$ npx skills add pytorch/executorch --skill binary-size -a claude-codeProject install by default; add -g for ~/.claude/skills/.
$ gh skill install pytorch/executorch binary-size --agent claude-codeProject scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).
$ git clone --depth 1 https://github.com/pytorch/executorch.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.claude/skills/binary-size .claude/skills/binary-size && rm -rf skills-srcUse ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.
Claude Code skills documentation · loads skills from .claude/skills/
Install the "binary-size" agent skill from https://github.com/pytorch/executorch/tree/main/.claude/skills/binary-size into .claude/skills/binary-size/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "binary-size", then confirm the skill loads.Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$skill-installer install https://github.com/pytorch/executorch/tree/main/.claude/skills/binary-sizeType this inside Codex. $skill-installer <name> installs a curated skill from openai/skills. The installer writes to $CODEX_HOME/skills (default ~/.codex/skills). Restart Codex if the skill does not show up.
$ npx skills add pytorch/executorch --skill binary-size -a codexProject install goes to .agents/skills/; add -g for ~/.codex/skills/.
$ gh skill install pytorch/executorch binary-size --agent codexProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/pytorch/executorch.git skills-src && mkdir -p .agents/skills && cp -r skills-src/.claude/skills/binary-size .agents/skills/binary-size && rm -rf skills-srcUse ~/.agents/skills/ instead of .agents/skills for a personal install.
Codex skills documentation · loads skills from .agents/skills/
Install the "binary-size" agent skill from https://github.com/pytorch/executorch/tree/main/.claude/skills/binary-size into .agents/skills/binary-size/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "binary-size", then confirm the skill loads.Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add pytorch/executorch --skill binary-size -a cursorProject install goes to .agents/skills/; add -g for ~/.cursor/skills/.
$ gh skill install pytorch/executorch binary-size --agent cursorProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/pytorch/executorch.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/.claude/skills/binary-size .cursor/skills/binary-size && rm -rf skills-srcUse ~/.cursor/skills/ instead of .cursor/skills for a personal install.
Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/
Install the "binary-size" agent skill from https://github.com/pytorch/executorch/tree/main/.claude/skills/binary-size into .cursor/skills/binary-size/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "binary-size", then confirm the skill loads.Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gemini skills install https://github.com/pytorch/executorch.git --path .claude/skills/binary-size--scope user (default) or --scope workspace; --path is the subfolder of the repo that holds the skill; --consent skips the security confirmation prompt.
$ npx skills add pytorch/executorch --skill binary-size -a gemini-cliProject install goes to .agents/skills/; add -g for ~/.gemini/skills/.
$ gh skill install pytorch/executorch binary-size --agent gemini-cliProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/pytorch/executorch.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/.claude/skills/binary-size .gemini/skills/binary-size && rm -rf skills-srcUse ~/.gemini/skills/ instead of .gemini/skills for a personal install, then run /skills reload.
Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/
Install the "binary-size" agent skill from https://github.com/pytorch/executorch/tree/main/.claude/skills/binary-size into .gemini/skills/binary-size/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "binary-size", then confirm the skill loads.Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gh skill install pytorch/executorch binary-sizeInstalls for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).
$ npx skills add pytorch/executorch --skill binary-size -a github-copilotProject install goes to .agents/skills/; add -g for ~/.copilot/skills/.
$ git clone --depth 1 https://github.com/pytorch/executorch.git skills-src && mkdir -p .github/skills && cp -r skills-src/.claude/skills/binary-size .github/skills/binary-size && rm -rf skills-srcUse ~/.copilot/skills/ instead of .github/skills for a personal install. Commit .github/skills so cloud agent and code review can use it.
GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/
Install the "binary-size" agent skill from https://github.com/pytorch/executorch/tree/main/.claude/skills/binary-size into .github/skills/binary-size/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "binary-size", then confirm the skill loads.GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add pytorch/executorch --skill binary-size -a opencodeOpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
$ gh skill install pytorch/executorch binary-size --agent opencodeProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/pytorch/executorch.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/.claude/skills/binary-size .opencode/skills/binary-size && rm -rf skills-srcUse ~/.config/opencode/skills/ instead of .opencode/skills for a personal install.
OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/
Install the "binary-size" agent skill from https://github.com/pytorch/executorch/tree/main/.claude/skills/binary-size into .opencode/skills/binary-size/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "binary-size", then confirm the skill loads.OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
Facts
- Skill name
binary-size- GitHub stars
- 5.1k
- Token cost
- ~793 tokens
- SKILL.md length
- 280 words
- Files
- 1
- Skills in repo
- 10
- Repo updated
- First seen
- Licence
- Custom licence
At a glance
Measures and shrinks the ExecuTorch runtime binary by building a size test, analyzing it with bloaty and landing each reduction as its own pull request.
- Works in 4 steps: Create a branch: git checkout -b… → Implement, rebuild, measure stripped sizes → Create a separate PR — one logical… → …
- Investigating why the ExecuTorch runtime binary grew between two builds
- SKILL.md covers Start from the main branch of…, Build and measure baseline, Analyze with bloaty and Key build flags, plus 3 more sections
- Calls git, conda and bash
What it does
The agent asks where the executorch repo lives, updates main, activates the executorch conda environment and runs the build size test script, then strips the resulting binary to get a baseline. Two binaries come out: size_test, the runtime without operator implementations, and size_test_all_ops, which includes portable ops. bloaty breaks them down by symbol or section and can diff two builds.
The skill lists where savings usually hide: large functions and template bloat in .text, verbose messages and file paths in .rodata, static initializers, and logging strings that disappear when ET_LOG_ENABLED is set to 0 in Release. Constraints are firm: build with CMake, require C++17, stay compatible with GCC 9 and Clang 12, keep latency unchanged, leave the size script's flags alone and run tests for touched files. Each change goes on its own branch and pull request, with results recorded in a markdown file.
When your agent uses it
- Investigating why the ExecuTorch runtime binary grew between two builds
- Looking for size savings for a size-constrained on-device deployment
- Running the size test and comparing stripped sizes before and after a change
Example prompts
- “Build the ExecuTorch size test and show me the 30 biggest symbols.”
- “Compare the stripped size_test binary before and after my branch.”
- “Find places where logging strings or static initializers add to the binary size.”
Requirements
- A local executorch checkout
- A conda environment named executorch
- CMake and the bloaty tool
Workflow steps
4 steps, taken from the first numbered list in SKILL.md.
- Create a branch: git checkout -b binary-size-
- Implement, rebuild, measure stripped sizes
- Create a separate PR — one logical change per PR
- Record results in binary-size-.md
What it can do on your machine
Read from SKILL.md and the folder at commit 27d124f. It shows what the files ask for, not the result of running them.
Tool permissions
Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.
From allowed-tools in the SKILL.md frontmatter.
Runs code
Shell commands in SKILL.md call:
gitcondabash
From the folder's file list and the shell code blocks in SKILL.md.
Network
No URLs in SKILL.md. Its commands use git, which can reach the network depending on how they are called.
From URLs in SKILL.md, links to its own repository left out.
Credentials
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.
Context cost
ExecuTorch Binary Size Reduction loads about 793 tokens when it runs. Until then it costs about 43 tokens; SKILL.md has 280 words of instructions outside code blocks.
Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.
Safety
Auto-check passed
The automated check found no risky patterns in SKILL.md.
Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.
SKILL.md
Its licence (Custom licence) doesn't allow us to republish the file, so here is its outline and opening line. It has 280 words (~793 tokens).
“Ask the user where the executorch repo is.”
- name
- binary-size
Files
Just SKILL.md in .claude/skills/binary-size of pytorch/executorch.
Open the folder on GitHubat commit 27d124f
Compare with similar skills
ExecuTorch Binary Size Reduction next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.
| Skill | Stars | Used in | Tokens | Auto-check | Licence | Repo updated |
|---|---|---|---|---|---|---|
| ExecuTorch Binary Size Reduction this skillpytorch/executorch | 5.1k | — | ~793 | Automated safety check: Pass | Custom licence | |
| The Art of Debuggingstas00/the-art-of-debugging | 1.7k | — | ~6.1k | Automated safety check: Notes | CC-BY-SA-4.0 | |
| Worktree Env Setupmeta-pytorch/attention-gym | 1.3k | — | ~858 | Automated safety check: Pass | BSD-3-Clause | |
| ONNX Opset Bump Checklistmicrosoft/onnxruntime | 22k | — | ~12k | Automated safety check: Pass | MIT | |
| Ascendcascend-ai-coding/awesome-ascend-skills | 174 | — | ~3.5k | Automated safety check: Pass | None | |
| External Mindstudio Ascend Profiler DB Explorerascend-ai-coding/awesome-ascend-skills | 174 | — | ~1.4k | Automated safety check: Pass | None |
Similar skills
Worktree Env Setup
meta-pytorch/attention-gym
Sets up an isolated per-worktree Python environment for attention-gym development using nightly PyTorch and the CI-mirroring uv flow.
- Official
ONNX Opset Bump Checklist
microsoft/onnxruntime
A checklist for upgrading the pinned ONNX version and opset in ONNX Runtime, covering the files to change, archive hashes, patch rebasing and release-candidate handling.
Ascendc
ascend-ai-coding/awesome-ascend-skills
End-to-end AscendC custom operator development for Ascend NPU in an ascend-kernel (csrc/ops + build.sh + torchnpu PyTorch custom op) project.
External Mindstudio Ascend Profiler DB Explorer
ascend-ai-coding/awesome-ascend-skills
面向 Ascend PyTorch Profiler / msprof DB（如 ascendpytorchprofiler.db、msprof.db）的 SQL 分析技能。将自然语言问题（算子耗时、通信、下发、调度、schema/table 查询）转为安全可执行 SQL，并按需从官方文档提取表结构详情。
Code Review Checklist
shareAI-lab/learn-claude-code
Reviews code against a five-part checklist covering security, correctness, performance, maintainability and testing, and reports findings in a fixed format.
More from pytorch/executorch
All 10 skills in this repoExecuTorch Build Guide
pytorch/executorch
Builds ExecuTorch from source: the Python package, C++ runtime, model runners, Android and iOS cross-compilation and backend-specific builds, with environment checks.
ExecuTorch Cortex-M Backend
pytorch/executorch
Developer guide for the Cortex-M (CMSIS-NN) backend in ExecuTorch: quantization pipeline, pass manager, tests and adding new ops.
ExecuTorch Knowledge Base
pytorch/executorch
Answers ExecuTorch questions from a local wiki on backends, export pitfalls, quantization recipes, runtime errors and SoC compatibility.
ExecuTorch PR Review
pytorch/executorch
Reviews ExecuTorch pull requests or local branches for what CI cannot check, using a checklist, with an optional detailed line-by-line mode.
Qualcomm QNN Backend Development
pytorch/executorch
Helps build, test and extend the Qualcomm AI Engine Direct (QNN) backend in ExecuTorch, with routes for new ops, model export, Buck-vs-CMake parity fixes and per-layer accuracy debugging.
ExecuTorch on Zephyr
pytorch/executorch
Sets up ExecuTorch as a Zephyr RTOS module, adds board support and debugs west build failures such as linker memory overflow on embedded boards.
Related
Categories
Questions about ExecuTorch Binary Size Reduction
What does ExecuTorch Binary Size Reduction do?
Measures and shrinks the ExecuTorch runtime binary by building a size test, analyzing it with bloaty and landing each reduction as its own pull request. The agent asks where the executorch repo lives, updates main, activates the executorch conda environment and runs the build size test script, then strips the resulting binary to get a baseline. Two binaries come out: size_test, the runtime without operator implementations, and size_test_all_ops, which includes portable ops.
When should I use ExecuTorch Binary Size Reduction?
ExecuTorch Binary Size Reduction fits situations like: investigating why the ExecuTorch runtime binary grew between two builds; looking for size savings for a size-constrained on-device deployment; running the size test and comparing stripped sizes before and after a change.
How do I install ExecuTorch Binary Size Reduction in Claude Code?
Run `npx skills add pytorch/executorch --skill binary-size -a claude-code`. Or copy the skill folder (.claude/skills/binary-size in pytorch/executorch) into .claude/skills/binary-size in your project. Claude Code loads it when a task matches its description.
How do I install ExecuTorch Binary Size Reduction in Codex?
Run `npx skills add pytorch/executorch --skill binary-size -a codex`. Or copy the skill folder (.claude/skills/binary-size in pytorch/executorch) into .agents/skills/binary-size in your project. Codex loads it when a task matches its description.
Can I use ExecuTorch Binary Size Reduction in Cursor, Gemini CLI or GitHub Copilot?
Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add pytorch/executorch --skill binary-size -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/binary-size, .gemini/skills/binary-size, .github/skills/binary-size and .opencode/skills/binary-size in your project.
What does ExecuTorch Binary Size Reduction need to run?
Going by SKILL.md and its folder, ExecuTorch Binary Size Reduction needs the command-line tools its instructions call (git, conda and bash). Our summary lists: A local executorch checkout; A conda environment named executorch; CMake and the bloaty tool.
Does ExecuTorch Binary Size Reduction access the network?
SKILL.md contains no URLs. Its commands use git, which can reach the network depending on how they are called. This is read from the text; nothing was executed.
Is ExecuTorch Binary Size Reduction safe to install?
Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.
What licence does ExecuTorch Binary Size Reduction use?
ExecuTorch Binary Size Reduction has a licence file (the repository's licence) that doesn't match a standard licence. Read it on GitHub before reusing the skill.
How many tokens does ExecuTorch Binary Size Reduction use?
About 793 tokens (SKILL.md is roughly 3.2k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.
What are the alternatives to ExecuTorch Binary Size Reduction?
Skills that share tags, products or a category with ExecuTorch Binary Size Reduction: The Art of Debugging (stas00/the-art-of-debugging, 1.7k stars), Worktree Env Setup (meta-pytorch/attention-gym, 1.3k stars), ONNX Opset Bump Checklist (microsoft/onnxruntime, 22k stars) and Ascendc (ascend-ai-coding/awesome-ascend-skills, 174 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
Who maintains ExecuTorch Binary Size Reduction?
pytorch (a GitHub organization) maintains it in pytorch/executorch, which has 5,081 GitHub stars. The repository holds 10 skills in this directory. The repository was last updated on October 7, 2026.
Source: pytorch/executorch on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.