Skip to content

feiskyer/claude-code-settings / skills/skill-creator

skill-creator

Verified

Create, refine, and benchmark agent skills. Use when building a new skill, updating an existing one, running evals, checking trigger quality, or improving a skill description.

FileKindSize
LICENSE.txtmarkdown11.1 KB
README.mdmarkdown462 B
SKILL.mdskill md30.0 KB
agents/analyzer.mdmarkdown10.4 KB
agents/comparator.mdmarkdown6.9 KB
agents/grader.mdmarkdown9.1 KB
assets/eval_review.htmlother6.9 KB
eval-viewer/generate_review.pyscript16.0 KB
eval-viewer/viewer.htmlother43.9 KB
references/schemas.mdmarkdown11.6 KB
scripts/__init__.pyscript0 B
scripts/aggregate_benchmark.pyexecutablescript14.5 KB
scripts/generate_report.pyexecutablescript13.0 KB
scripts/improve_description.pyscript16.2 KB
scripts/package_skill.pyexecutablescript4.1 KB
scripts/platforms/__init__.pyscript54 B
scripts/platforms/base.pyscript2.2 KB
scripts/platforms/claude_code.pyscript6.3 KB
scripts/platforms/codex.pyscript5.4 KB
scripts/platforms/openclaw.pyscript4.3 KB
scripts/quick_validate.pyscript9.6 KB
scripts/run_eval.pyexecutablescript8.2 KB
scripts/run_loop.pyexecutablescript15.0 KB
scripts/utils.pyscript3.4 KB

scripts/platforms/base.py

View on GitHub
"""Abstract base classes for platform adapters used in trigger evaluation."""

from abc import ABC, abstractmethod
from dataclasses import dataclass, field
from pathlib import Path
from typing import Any


@dataclass
class RegisteredSkill:
    """A temporary skill registered for one trigger-eval run."""

    name: str
    detection_token: str
    detection_mode: str
    detection_details: str
    cleanup_paths: list[Path] = field(default_factory=list)


@dataclass
class QueryRunResult:
    """Structured result from a platform query execution."""

    response_text: str
    raw_output: str
    metadata: dict[str, Any] = field(default_factory=dict)


class PlatformAdapter(ABC):
    """Defines the interface for running trigger evaluations on a specific platform."""

    @abstractmethod
    def find_project_root(self) -> Path:
        """Return the platform's project root directory."""
        ...

    @abstractmethod
    def register_skill(self, skill_name: str, description: str, sentinel: str) -> RegisteredSkill:
        """Register a temporary skill for testing."""
        ...

    def unregister_skill(self, registration: RegisteredSkill) -> None:
        """Remove any paths created by register_skill()."""
        for path in sorted(registration.cleanup_paths, key=lambda item: len(item.parts), reverse=True):
            if path.is_file():
                path.unlink(missing_ok=True)
                continue
            if path.is_dir():
                try:
                    path.rmdir()
                except OSError:
                    # Ignore non-empty directories created outside this run.
                    pass

    @abstractmethod
    def run_query(self, query: str, timeout: int, model: str | None) -> QueryRunResult | None:
        """Run a query and return the structured execution result.

        Returns None on failure or timeout.
        """
        ...

    def detect_trigger(self, run_result: QueryRunResult, registration: RegisteredSkill) -> bool:
        """Return True if the skill was triggered.

        The default implementation is a heuristic that checks for a sentinel string
        in the assistant's response text.
        """
        return registration.detection_token in run_result.response_text
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 467kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 319kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 319kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 319kUpdated Claude CodeCodexVerifiedInstall