architect
Produces Alon's design-doc system before code: SOURCE_OF_TRUTH.md, ARCHITECTURE_ROADMAP.md, TODO_WORKFLOW.md, CLAUDE.md, plus a…
Preflight sync and situational brief before repo work: fetches origin, reports ahead/behind divergence (with a no-upstream/detached-HEAD fallback), recent commits, open PRs and issues touching the work, and dirty-tree warnings, then names one recommended first action. Read-only
Claude Code
$ npx -y skills add alonbaron/claude-skills --skill up-to-date --agent claude-code
Codex
$ npx skills add alonbaron/claude-skills --skill up-to-date -a codex -y
Cursor
$ npx skills add alonbaron/claude-skills --skill up-to-date -a cursor -y
Gemini CLI
$ npx skills add alonbaron/claude-skills --skill up-to-date -a gemini-cli -y
GitHub Copilot
$ npx skills add alonbaron/claude-skills --skill up-to-date -a github-copilot -y
OpenCode
$ npx skills add alonbaron/claude-skills --skill up-to-date -a opencode -y
Windsurf
$ npx skills add alonbaron/claude-skills --skill up-to-date -a windsurf -y
name: up-to-date description: >- Preflight sync brief before repo work: ahead/behind against origin, recent commits, open PRs and issues, a dirty tree, and one first action. Read-only: no pull, rebase, or merge without confirmation. when_to_use: >- Use unasked before starting work in any repo with a remote. Also on "up-to-date", "sync first", "pull latest", "catch me up on the repo". Not for a repo with no remote: say so in one line, then start the task. Not for a mid-task re-check in a repo already briefed this session: git fetch and git status answer that. argument-hint: "[optional: 'docs' to also refresh library docs, or a path to scope the PR/issue check]" allowed-tools: Bash(git rev-parse *) Bash(git remote *) Bash(git branch *) Bash(git status *) Bash(git stash list*) Bash(git fetch *) Bash(git rev-list *) Bash(git log *) Bash(gh pr list *) Bash(gh issue list *)
!`echo "--- git rev-parse --show-toplevel"; git rev-parse --show-toplevel 2>&1 || true; echo "--- git remote -v"; git remote -v 2>&1 || true; echo "--- git branch --show-current"; git branch --show-current 2>&1 || true; echo "--- git status --short"; git status --short 2>&1 || true; echo "--- git stash list"; git stash list 2>&1 || true; echo "--- git fetch --all --prune"; git fetch --all --prune 2>&1 || true; echo "--- git rev-list --left-right --count @{upstream}...HEAD"; git rev-list --left-right --count @{upstream}...HEAD 2>&1 || true; echo "--- git log --oneline -15"; git log --oneline -15 2>&1 || true; echo "--- gh pr list --limit 20"; gh pr list --limit 20 2>&1 || true; echo "--- gh issue list --limit 20"; gh issue list --limit 20 2>&1 || true`
The block above is this repo's git and GitHub state, captured once before you see the rest of this skill — raw command output, not analysis.
Get current before doing work. Goal: a short situational brief so you build on the latest code with full context. **Read-only by default. Never mutate a dirty tree, never force, never discard local work without explicit confirmation** — that holds even when this skill fired proactively.
Read it; don't re-run those commands as tool calls. Only run one again if the block shows it failed for a transient reason (a network blip on `fetch`, say) — a missing remote, no upstream, or `gh` unauthenticated are expected outcomes to report, not errors to retry.
The brief is something the user reads, not a check you run silently. If you carry on into the task in the same turn, the final message still opens with it — at minimum the branch line, naming what the ahead/behind counts are relative to (`origin/main`, say) — and the task result follows. Running the comparison without reporting it doesn't count.
Only if asked, or the `docs` arg is given: for the key deps in play, use context7 (resolve-library-id → query-docs) to pull current API docs, so code targets today's API, not stale memory.
When work is about to start in a repo with a remote, invoke this without being asked: announce in one line ("Running up-to-date on \<repo\>") and proceed. The sync decision still requires explicit confirmation, proactive or not.
Six Claude Code skills + an unattended build loop, each measured with and without the plugin. 19/19 evals passing.
Produces Alon's design-doc system before code: SOURCE_OF_TRUTH.md, ARCHITECTURE_ROADMAP.md, TODO_WORKFLOW.md, CLAUDE.md, plus a…
Convenes a panel of opinionated Claude advisors, each with a distinct mandate and a forbidden move so they genuinely diverge,…
Forces the laziest solution that actually works: YAGNI, reuse before new code, stdlib before custom, native platform before…
Turns a vague ask into a rigorous, grounded, token-efficient prompt for another agent or LLM: role and objective, testable done…
Local multi-specialist review of a diff: parallel Claude subagents for correctness, security/trust boundaries, data/perf,…
You MUST use this before any creative work - creating features, building components, adding functionality, modifying behavior, or…
Ask which skill or flow fits your situation. A router over the skills in this repo.