00 Start Here
by AIDevGTM
Interview the user once and write a docs/gtm-cofounder/founder-brief.md that every other skill reads first, so the advice is about their real business, not a...
Loading Skills Catalog: Organizing 2,800+ verified agent skills, categories, and live metrics...
Debug a MAX model that loads, compiles, serves, and generates tokens but whose output disagrees with a reference implementation. Use whenever parity debuggin...
Debug Model is an open-source AI agent skill designed for Claude Code, Codex CLI, Cursor, Gemini CLI. It enables autonomous agents to debug a max model that loads, compiles, serves, and generates tokens but whose output disagrees with a reference implementation. use whenever parity debuggin... with verified execution instructions and live repository tracking.
Pick the AI assistant you use, then copy the command or follow the steps. New to this? Read the install guide.
Security Notice: Listed from the author's public repository. Not owned or endorsed by SkillHub. Review before installing. Always inspect a skill's SKILL.md and execution scripts before running commands locally.
Installation Time: Downloading and configuring this skill typically takes 15β45 seconds. If your terminal pauses during package downloads, Git checkouts, or dependency resolution, please allow it to finish.
npx skills add modular/skills --skill debug-model
This skill uses a direct npx package distribution. Run the verified command below in your terminal:
npx skills add modular/skills --skill debug-model
debug-modelDebug Model β Debug a MAX model that loads, compiles, serves, and generates tokens but whose output disagrees with a reference implementation. Use whenever parity debugging stalls on scalar taps, the model returns gibberish or wrong greedy tokens, logit cosine is high but argmax differs, or generation is coherent then diverges. Applies during an architecture port, a quantization bring-up, a multi-GPU conversion, or after a MAX upgrade. Triggers on "parity failure", "silent corruption", "logits match but tokens diverge", "top-1 mismatch", "greedy divergence", and "model serves but generates garbage". Not for crashes on load or pre-serve scaffolding (use import-model). Mandates reference-vs-MAX tensor-dump comparators first, numerical verification of fixes before recompiling, and a serve-vs-pipeline bisect when dumps match but text diverges.
Install with npx skills add modular/skills --skill debug-model.
Source: debug-model in modular/skills.
The same install steps as plain text, for AI assistants to read:
/skills/debug-model/install.txtBroken link, old instructions or a safety worry? Tell us.
Report broken skill or link βCommon questions about installing and using this skill.
You can install Debug Model by cloning the repository or copying the skill instructions from our Install Tabs into your agent's skills directory (~/.claude/skills/debug-model/ for Claude Code or your project's .claude/ folder).
Debug Model officially supports Claude Code, Codex CLI, Cursor, Gemini CLI. Setup commands and agent prompts are pre-configured for each platform.
The official repository is hosted on GitHub at https://github.com/modular/skills by author modular under the MIT license.
by AIDevGTM
Interview the user once and write a docs/gtm-cofounder/founder-brief.md that every other skill reads first, so the advice is about their real business, not a...
by AIDevGTM
After the founder brief and the market research, turn them into an honest diagnosis and a prioritized, stage-aware GTM roadmap saved as docs/gtm-cofounder/gt...
by edonadei
Send an operational alert. Use when asked to send an alert, page, or notification.
by AIDevGTM
Define a real ICP and the developer personas in the sale. Use when the user says the product is \"for developers,\" can't name who would say no, or is market...