用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt;运行前请先检查来源。
npx skills add https://github.com/tinh2/skills-hub-registry --skill pr命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地?可下载 SkillsMP 当前能够提供的文件。
Safely connect X (formerly Twitter) data to Claude Code, Codex, Cursor, or Windsurf through MCP. Preserve existing providers, compare the official self-hosted XMCP server with an opt-in hosted Xquik connection, disclose data flow, and run bounded tweet search, trend, competitor, or bookmark workflows.
Raise a site's perceived production value with a rigorous UI/UX pass while PROVING no user-facing word changed. Use when the user says "make it look expensive", "premium design pass", "make it feel designed", "rigorous design audit", "improve visual craft", "brand-safe redesign", "polish the UI without changing copy", or "design pass but do not touch the words". ALSO use whenever a visual change is requested on a site whose copy was written or approved by someone else — a client, a lawyer, a clinician, a compliance team — because the real risk there is not ugly CSS, it is a heading getting quietly "improved" during a refactor. Do NOT use for greenfield design (that is a build skill) or for copywriting.
Autonomous API effectiveness evaluation — contract clarity, consumer ergonomics, error experience, evolvability, and craft quality. Produces actionable feedback with specific before/after recommendations. The API counterpart to /design-critique. Use when asked to "critique my API", "is my API well designed", "review API developer experience", "score my API", "how usable is this API", "API DX audit".
基于 SOC 职业分类
正在显示 SKILL.md
| name | pr |
| description | Create a convention-compliant pull request from the current branch. |
| version | 2.0.0 |
| category | review |
| platforms | ["CLAUDE_CODE"] |
You are a PR creation agent. Create a clean, convention-compliant pull request. Do NOT ask the user questions. Infer everything from git history and code changes.
INPUT: $ARGUMENTS (optional) Additional context or flags:
--draft or -d — create as draft PR--reviewer <user> or -r <user> — assign reviewer(s), comma-separatedgit branch --show-currentgit symbolic-ref refs/remotes/origin/HEAD -> strip refs/remotes/origin/develop exists, then main, then mastergit log {base}..HEAD --format="%s" --reversegit diff {base}..HEAD --statgit diff {base}..HEADExtract issue/story identifiers from the branch name using these patterns:
Jira-style (most common):
[A-Z][A-Z0-9]+-\d+ (e.g., DEV-4979, STORY-123, PROJ-42)DEV-4979-add-email-verification -> DEV-4979Linear-style:
[A-Z][A-Z0-9]+-\d+ (same as Jira, e.g., ENG-123, FE-45)eng-123-fix-auth -> ENG-123GitHub Issues:
(\d+)- at the start, or -(\d+)- after a prefix like fix/, feat/fix/42-broken-login -> #42, 123-add-feature -> #123If an identifier is found, determine the tracker type by checking these in order:
.pr-config or .github/pr-config.yml file in the repo root containing tracker settings (see CONFIGURATION below).github.com, and the identifier is purely numeric, assume GitHub Issues.[A-Z]+-\d+, assume Jira/Linear style. Build the URL from the configured base URL (see CONFIGURATION).The skill reads optional configuration from these locations (first match wins):
.pr-config.yml or .github/pr-config.yml in the repo root~/.config/claude-pr/config.ymlConfig schema:
# Issue tracker settings
tracker:
# Type: "jira", "linear", "github", or "none"
type: jira
# Base URL for building issue links (Jira/Linear)
url: https://myteam.atlassian.net/browse
# For Linear: https://linear.app/myteam/issue
# Deploy convention — a string to check in the last commit message
# Set to null or omit to skip this check entirely
deploy_tag: "deploy:username"
# Default reviewers (GitHub usernames)
reviewers: []
If no config file exists, use these defaults:
tracker.type: auto-detect from branch name and remotetracker.url: for Jira, attempt to read from any atlassian.net references in the repo; otherwise omit the linkdeploy_tag: null (skip check)reviewers: [] (none)Determine the change type from commit messages and diff:
feat: -> New featurefix: -> Bug fixrefactor: -> Code restructuredocs: -> Documentationtest: -> Test changeschore: -> MaintenanceUse the most common prefix across commits, or the most significant change type.
Title (under 70 characters):
{type}: ({STORY-NUMBER}) {brief description}{type}: {brief description}Body:
## Summary
{2-4 bullet points describing what changed and why -- focus on the "why"}
## Changes
{List key files changed with brief explanation of each change}
## Test Plan
- [ ] {Specific testable verification step}
- [ ] {Another verification step}
- [ ] All existing tests pass
## Issue
{Link to the issue, formatted based on tracker type:}
{Jira: [DEV-4979](https://myteam.atlassian.net/browse/DEV-4979)}
{Linear: [ENG-123](https://linear.app/myteam/issue/ENG-123)}
{GitHub: Closes #42}
If no story/issue number was detected, omit the Issue section entirely. If tracker URL is not configured, just show the identifier without a link.
git rev-parse --verify origin/{branch} 2>/dev/null
If not pushed: git push -u origin {branch}gh pr view {branch} --json number 2>/dev/null
If exists: update it with gh pr edit instead of creating new.deploy_tag is configured (non-null), verify last commit message contains it.
If not, warn the user but still create the PR.Build the gh pr create command:
gh pr create --title "{title}" --body "{body}" --base {base-branch}
Add flags based on input and config:
--draft was passed in $ARGUMENTS: add --draft--reviewer was passed or reviewers is configured: add --reviewer {user1} --reviewer {user2}Use a HEREDOC for the body to preserve formatting.
If updating an existing PR:
gh pr edit {number} --title "{title}" --body "{body}"
STRICT CONVENTIONS (from CLAUDE.md):
OUTPUT:
After producing the review, validate completeness and consistency:
IF VALIDATION FAILS:
After producing output, record execution metadata for the /evolve pipeline.
Check if a project memory directory exists:
~/.claude/projects/skill-telemetry.md in that memory directoryEntry format:
### /pr — {{YYYY-MM-DD}}
- Outcome: {{SUCCESS | PARTIAL | FAILED}}
- Self-healed: {{yes — what was healed | no}}
- Iterations used: {{N}} / {{N max}}
- Bottleneck: {{phase that struggled or "none"}}
- Suggestion: {{one-line improvement idea for /evolve, or "none"}}
Only log if the memory directory exists. Skip silently if not found. Keep entries concise — /evolve will parse these for skill improvement signals.