Перейти к содержимому
TakoAPI
Назад к навыкам

code-cache

от ryx2

Search & Research
487Число загрузок по данным clawskills.sh
0Просмотры
0 лайковВойдите, чтобы поставить лайк навыку
—Ещё нет оценок
Ваша оценка

Кратко

Semantic code caching for AI agents.

Описание

Что делает этот навык

Stores code generated during agent executions and retrieves semantically similar code for future tasks. Uses the Raysurfer API to search, upload, and vote on cached snippets. Cuts redundant generation by reusing validated solutions from prior runs.

Retrieving pre-validated cached code avoids the latency and token cost of having an LLM regenerate the same patterns on every run.

Когда его использовать

  • Reusing a prior data pipeline script without regenerating it
  • Retrieving cached chart-building code for a recurring report
  • Searching past executions before writing new API fetch code
  • Uploading a successful script to the cache after an agent run
  • Downvoting a broken cached snippet to improve future retrieval quality

Пример сценария

ВВОД

User asks: 'Fetch GitHub trending repos'

АГЕНТ
  1. 1

    Search cache with /code-cache search "Fetch GitHub trending repos" --show-code

  2. 2

    Review the top match score and inspect the returned snippet

  3. 3

    Retrieve execution-ready files with /code-cache files "Fetch GitHub trending repos"

  4. 4

    Run the retrieved code in the agent sandbox

  5. 5

    Upload the result with /code-cache upload "Fetch GitHub trending repos" --files fetch.py

ВЫВОД

GitHub trending repos fetched using cached code, with the execution stored and auto-voted for future reuse

Требования

  • RAYSURFER_API_KEY environment variable
  • Raysurfer account (sign up at raysurfer.com to access the API key dashboard)

Установка

clawhub install ryx2/code-cache

Выполните команду CLI или вставьте промпт в Claude Code, Cursor, Codex — любой кодинг-агент, которым пользуетесь.

Ссылки