Agent skill

Testing Fwd Bwd Kernels

by sablin39 in sablin39/tilelang-cuda-skills

Implement and validate forward and backward TileLang kernels integrated with torch.autograd.Function.

No licenceAuto-check passedDevelopment

Install Testing Fwd Bwd Kernels

skills CLI
$ npx skills add sablin39/tilelang-cuda-skills --skill testing-fwd-bwd-kernels -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install sablin39/tilelang-cuda-skills testing-fwd-bwd-kernels --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/sablin39/tilelang-cuda-skills.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/tilelang/testing-fwd-bwd-kernels .claude/skills/testing-fwd-bwd-kernels && rm -rf skills-src

Use ~/.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/

Facts

Skill name
testing-fwd-bwd-kernels
GitHub stars
145
Token cost
~1.2k tokens
SKILL.md length
530 words
Files
3 (incl. references)
Skills in repo
10
Repo updated
First seen
Licence
None found

At a glance

Implement and validate forward and backward TileLang kernels integrated with torch.autograd.Function.

  • Works in 4 steps: Run custom and reference paths on… → Use the same random upstream gradient… → Compare each differentiable input's… → …
  • Custom differentiable operators and gradient mismatches
  • SKILL.md covers Define the contract, Integrate with autograd, Validate independently and Localize a mismatch, plus 1 more section
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md

What it does

Testing Fwd Bwd Kernels is an agent skill from sablin39/tilelang-cuda-skills. Implement and validate forward and backward TileLang kernels integrated with torch.autograd.Function. Use for custom differentiable operators and gradient mismatches; use the writing or debugging skills for forward-only kernels.

Its SKILL.md is about 1.2k tokens, which your agent loads only when the skill is triggered. The skill folder holds 3 other files, including reference files (for example `references/fwd-bwd-examples.md` and `references/gradient-testing-patterns.md`).

It sits in Development. The repository describes itself as: Skills for writing tilelang and debugging with CUDA toolkits.

When your agent uses it

  • Custom differentiable operators and gradient mismatches
  • Use the writing
  • Debugging skills for forward-only kernels

Example prompts

  • “/testing-fwd-bwd-kernels”

Workflow steps

4 steps, taken from the first numbered list in SKILL.md.

  1. Run custom and reference paths on detached input clones with matching
  2. Use the same random upstream gradient for both paths. torch.autograd.grad
  3. Compare each differentiable input's gradient, including missing gradients.
  4. Cover rectangular and tail shapes, supported dtypes, masks/causal modes, and

What it can do on your machine

Read from SKILL.md and the folder at commit 502a514. 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

    No scripts in the folder and no shell commands in SKILL.md.

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    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

Testing Fwd Bwd Kernels loads about 1.2k tokens when it runs, and up to ~3.8k if it reads all its reference files. Until then it costs about 63 tokens; SKILL.md has 530 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~63
When it runs · the whole SKILL.md, loaded when a task matches
~1.2k
With references · SKILL.md plus every file in references/, read only if the agent opens them
~3.8k

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

Without a licence we can't republish the file, so here is its outline and opening line. It has 530 words (~1,167 tokens).

“For performance work on both phases, use the joint optimization loop. Validate the forward/backward pair and saved-state contract together; select using the task's combined objective and record per-phase tradeoffs.”

— opening of SKILL.md by sablin39
name
testing-fwd-bwd-kernels

Read the full SKILL.md on GitHub

Files

SKILL.md and 2 other files (references) in skills/tilelang/testing-fwd-bwd-kernels of sablin39/tilelang-cuda-skills.

  • SKILL.md
  • references/fwd-bwd-examples.md
  • references/gradient-testing-patterns.md

Open the folder on GitHubat commit 502a514

Compare with similar skills

Testing Fwd Bwd Kernels 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.

Testing Fwd Bwd Kernels compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Testing Fwd Bwd Kernels this skillsablin39/tilelang-cuda-skills145—~1.2kAutomated safety check: PassNone
Vercel Composition Patternssupabase/supabase111k58 repos~726Automated safety check: PassMIT
Finishing a Development Branchobra/superpowers297k5 repos~1.9kAutomated safety check: PassMIT
Typescript Advanced Typesrolling-scopes/rsschool-app10k25 repos~4.2kAutomated safety check: PassMPL-2.0
PR Babysitteropeninterpreter/openinterpreter69k3 repos~4.2kAutomated safety check: PassApache-2.0
Code Review ChecklistshareAI-lab/learn-claude-code78k4 repos~1.1kAutomated safety check: PassMIT

Similar skills

  • Official

    React composition patterns that scale. An agent skill from supabase/supabase.

    111k GitHub starsUsed in 58 repos~726 tokens
    DevelopmentAuto-check passed
  • Walks the last step of a branch: confirm tests pass, detect the git environment, ask how to integrate, carry out your choice and clean up the worktree.

    297k GitHub starsUsed in 5 repos~1.9k tokens
    DevelopmentAuto-check passed
  • Typescript Advanced Types

    rolling-scopes/rsschool-app

    Master TypeScript's advanced type system including generics, conditional types, mapped types, template literals, and utility types for building type-safe applications.

    10k GitHub starsUsed in 25 repos~4.2k tokens
    DevelopmentAuto-check passed
  • PR Babysitter

    openinterpreter/openinterpreter

    Watches an open GitHub pull request until it merges, handling review comments, diagnosing CI failures and retrying flaky checks along the way.

    69k GitHub starsUsed in 3 repos~4.2k tokens
    DevelopmentAuto-check passed
  • 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.

    78k GitHub starsUsed in 4 repos~1.1k tokens
    DevelopmentAuto-check passed
  • Greploop

    onyx-dot-app/onyx

    Iteratively improves a PR (GitHub), MR (GitLab), or shelved changelist (Perforce) until Greptile gives it a 5/5 confidence score with zero unresolved comments.

    32k GitHub starsUsed in 4 repos~3.3k tokens
    DevelopmentAuto-check passed

More from sablin39/tilelang-cuda-skills

All 10 skills in this repo
  • Torch Profiling Tilelang Programs

    sablin39/tilelang-cuda-skills

    Inspect a PyTorch baseline with torch.compile graphs/code and torch.profiler before drafting a custom kernel, then compare TileLang and PyTorch timelines, forward/backward regions, and allocations.

    145 GitHub stars~1.3k tokensUpdated 25 days ago
    Auto-check passed
  • Writing Tilelang Design Docs

    sablin39/tilelang-cuda-skills

    Write human-facing TileLang design documents for proposed or completed kernel implementations.

    145 GitHub stars~2.2k tokensUpdated 25 days ago
    Auto-check passed
  • Cuda

    sablin39/tilelang-cuda-skills

    Draft, debug, and measure CUDA kernels and host launch workflows.

    145 GitHub stars~990 tokensUpdated 25 days ago
    Auto-check passed
  • Debugging Tilelang Programs

    sablin39/tilelang-cuda-skills

    Diagnose TileLang compilation, host argument, memory-liveness, runtime, and numerical failures.

    145 GitHub stars~1.6k tokensUpdated 25 days ago
    Auto-check passed
  • Designing With Tilefoundry

    sablin39/tilelang-cuda-skills

    Describe model or operator HIR with TileFoundry, inspect topology and placement, run static cost analysis, visualize authored dataflow, and check a runtime implementation against its evaluator.

    145 GitHub stars~3.4k tokensUpdated 25 days ago
    Auto-check passed
  • Profiling Tilelang Programs

    sablin39/tilelang-cuda-skills

    Measure TileLang kernel and workflow performance, choose benchmark or timeline tools, and compare implementations fairly.

    145 GitHub stars~1.6k tokensUpdated 25 days ago
    Auto-check passed

Categories

Questions about Testing Fwd Bwd Kernels

What does Testing Fwd Bwd Kernels do?

Implement and validate forward and backward TileLang kernels integrated with torch.autograd.Function. Testing Fwd Bwd Kernels is an agent skill from sablin39/tilelang-cuda-skills.Function.

When should I use Testing Fwd Bwd Kernels?

Testing Fwd Bwd Kernels fits situations like: custom differentiable operators and gradient mismatches; use the writing; debugging skills for forward-only kernels.

How do I install Testing Fwd Bwd Kernels in Claude Code?

Run `npx skills add sablin39/tilelang-cuda-skills --skill testing-fwd-bwd-kernels -a claude-code`. Or copy the skill folder (skills/tilelang/testing-fwd-bwd-kernels in sablin39/tilelang-cuda-skills) into .claude/skills/testing-fwd-bwd-kernels in your project. Claude Code loads it when a task matches its description.

How do I install Testing Fwd Bwd Kernels in Codex?

Run `npx skills add sablin39/tilelang-cuda-skills --skill testing-fwd-bwd-kernels -a codex`. Or copy the skill folder (skills/tilelang/testing-fwd-bwd-kernels in sablin39/tilelang-cuda-skills) into .agents/skills/testing-fwd-bwd-kernels in your project. Codex loads it when a task matches its description.

Can I use Testing Fwd Bwd Kernels 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 sablin39/tilelang-cuda-skills --skill testing-fwd-bwd-kernels -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/testing-fwd-bwd-kernels, .gemini/skills/testing-fwd-bwd-kernels, .github/skills/testing-fwd-bwd-kernels and .opencode/skills/testing-fwd-bwd-kernels in your project.

What does Testing Fwd Bwd Kernels need to run?

SKILL.md names no scripts, command-line tools or credentials: Testing Fwd Bwd Kernels is instructions for the agent only.

Does Testing Fwd Bwd Kernels access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Testing Fwd Bwd Kernels 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 Testing Fwd Bwd Kernels use?

No licence was found for Testing Fwd Bwd Kernels or its repository. Without one, default copyright applies: ask the author before reusing or redistributing it.

How many tokens does Testing Fwd Bwd Kernels use?

About 1.2k tokens (SKILL.md is roughly 4.7k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 2.7k tokens, read only when the agent opens those files.

What are the alternatives to Testing Fwd Bwd Kernels?

Skills that share tags, products or a category with Testing Fwd Bwd Kernels: Vercel Composition Patterns (supabase/supabase, 111k stars), Finishing a Development Branch (obra/superpowers, 297k stars), Typescript Advanced Types (rolling-scopes/rsschool-app, 10k stars) and PR Babysitter (openinterpreter/openinterpreter, 69k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Testing Fwd Bwd Kernels?

sablin39 (a GitHub user) maintains it in sablin39/tilelang-cuda-skills, which has 145 GitHub stars. The repository holds 10 skills in this directory. The repository was last updated on September 15, 2026.

Source: sablin39/tilelang-cuda-skills on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.