Profile My skills Saved My plugins Edit profile Submit a skill

← all publishers

sayamsingla2000

@sayamsingla2000 source repo

5 published skills

  1. ▌
    Grill · sayamsingla2000
    Interview the user relentlessly about a plan or coding task until reaching shared understanding at the level of files, signatures, and edge cases, resolving each branch of the decision tree. Use when user wants to stress-test a plan, get grilled on their design, align before implementation, or mentions "grill me".
    0 installs
  2. ▌
    Bootstrap · sayamsingla2000 bundle
    Bootstrap a new Python project with a working pytest harness, git repo, Makefile, and working-agreement AGENTS.md. Use when the user wants to start a new project, scaffold a repo, set up a fresh codebase, get a test loop running from scratch, or says "new project" / "bootstrap" / "scaffold this".
    0 installs
  3. ▌
    Tdd Workflow · sayamsingla2000
    Use this skill when writing new features, fixing bugs, or refactoring code. Enforces test-driven development with 80%+ coverage including unit, integration, and E2E tests.
    0 installs
  4. ▌
    Prompt Optimizer · sayamsingla2000
    Analyze raw prompts, identify intent and gaps, match ECC components (skills/commands/agents/hooks), and output a ready-to-paste optimized prompt. Advisory role only — never executes the task itself. TRIGGER when: user says "optimize prompt", "improve my prompt", "how to write a prompt for", "help me prompt", "rewrite this prompt", or explicitly asks to enhance prompt quality. Also triggers on Chinese equivalents: "优化prompt", "改进prompt", "怎么写prompt", "帮我优化这个指令". DO NOT TRIGGER when: user wants the task executed directly, or says "just do it" / "直接做". DO NOT TRIGGER when user says "优化代码", "优化性能", "optimize performance", "optimize this code" — those are refactoring/performance tasks, not prompt optimization.
    0 installs
  5. ▌
    Prompt Optimization Loop · sayamsingla2000 bundle
    Tune a user's own LLM prompt when its output doesn't match the required result — drive revisions from observed failures and feedback instead of rewording blind. Works for any prompt type: classification/LLM-judge, structured extraction, query or code generation, summarization, free-text generation. Covers a quick manual pass (a few bad outputs, revise now) and a full automated loop (eval set, held-out scoring, early stopping, multiple starts), and keeps a per-prompt record so a session months later resumes instead of restarting. Use when a prompt isn't producing the needed result, when wrong-vs-expected outputs are available for it, or when a prompt needs systematic improvement against an eval set.
    0 installs