원클릭으로
reflect
Review recent learnings and open PRs to improve skill files
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
메뉴
Review recent learnings and open PRs to improve skill files
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Search older daily logs and memory files for past context via npx qmd
Create a GitHub issue from a Slack thread conversation. Analyzes the discussion, extracts key points, searches for related issues, and creates a well-structured issue.
Aliases for commonly used repositories. The agent can interact with any public GitHub repo, but these aliases provide shortcuts.
Review, audit, or summarize a GitHub pull request — fetch the diff, check for bugs/security/test/contract issues, post inline comments or approve/request-changes via `gh`, and report back to Slack. Use this whenever the user asks to review, look at, check, audit, summarize, "take a pass at", or give feedback on a PR (or shares a github.com/.../pull/<n> URL) — even if they don't use the word "review" explicitly.
Create a new runtime skill that persists across deploys. Use when the user asks to add a new capability, automate a workflow, or teach the bot a new behavior. Also use when the user says "learn how to..." or "remember how to..." for repeatable procedures.
General GitHub operations using the gh CLI. Search issues, read issue details, find pull requests, and manage labels.
| name | reflect |
| description | Review recent learnings and open PRs to improve skill files |
Use this skill when asked to review recent learnings or improve your own skills.
Use the memory-search skill to find recent patterns and learnings, or read the last 7 days of daily logs directly from the memory base directory provided in context:
ls -t <memory_base_dir>/shared/daily/ | head -7
Read each file and identify:
Update shared/MEMORY.md if needed — consolidate, don't just append.
When asked to improve your skills, or when you notice recurring patterns that a skill update would fix:
memory-search skill — what keeps coming up? What mistakes repeat?github skill's PR workflowskills/ or prompts/system.md — add missing context, fix incorrect guidancegithub skill — keep changes small and focused, one theme per PRskills/**/*.md — skill content and instructionsprompts/system.md — system prompt tweaksMEMORY.md — if committing shared knowledge to the repo makes sensesrc/** — no code changes, only markdown