Great use cases
Pace skills get powerful when you layer them with Claude Code's broader features: Routines that run on a schedule, fire on GitHub events, or trigger from an API call; hooks that react to session lifecycle events; sub-agents that work in parallel; plan mode for review-before-act; and MCP connectors that read your real data. This page is the cookbook: one section per team, real workflows you can copy.
Sales
Pipeline brief on Monday. Account-prep when a meeting hits the calendar. Follow-ups drafted from the call transcript.
Monday morning pipeline brief, in your inbox by 7am
A scheduled job kicks off /sales:pipeline-review against HubSpot, slices by stage, and spawns one sub-agent per top-10 account to run /sales:account-research. The final brief lands in #sales-pipeline on Slack before standup. You walk in with the report already read.
Auto-prep when a meeting lands on the calendar
The hook reads the attendee's email, runs /sales:call-prep, and drops the brief into the event description. By the time you open the meeting, the brief is already there: account history, recent activity, talking points, three questions to ask.
Draft follow-ups straight from the Zoom transcript
The hook pulls the transcript, runs /sales:call-summary to extract decisions and action items, then /sales:draft-outreach to compose the follow-up. Plan mode parks the draft for your approval. One click sends; the reply gets logged back to HubSpot, and /apollo:sequence-load can enroll the contact into a multi-touch sequence if the deal warrants it.
Marketing
Weekly campaign brief. Content drafting that knows your brand voice. Performance reports you don't have to assemble.
Weekly campaign performance digest
Friday afternoon, a scheduled job runs /marketing:performance-report against last week's Amplitude funnel + HubSpot lead source data, formats it as a Slack-friendly summary, and posts to #marketing-weekly. The team walks into Monday already aligned on what shipped and how it performed.
Brand-voice gate on every external draft
Any PR that touches landing pages, blog posts, or email templates auto-runs /brand-voice:brand-voice-enforcement and posts inline comments with concrete fixes. Catches drift before it ships; pair with /marketing:draft-content when a section needs a rewrite from scratch.
Engineering
Spec → stacked PRs → review → ship, with disciplined diagnosis for hard bugs and a careful-review pass before merge. The engineering plugin's 34 skills cover the whole arc.
Spec a feature, ship it as stacked PRs
/engineering:spec walks the feature into a parent epic with ordered sub-issues, one PR each. /engineering:implement builds them in stacked branches, opening one stacked PR per sub-issue. As reviews come in, /engineering:review handles the comment queue. When an upstream PR merges, /engineering:topr rebases the next one cleanly (correctly dropping commits that squash-merged). /engineering:next resets for the next task. The whole arc lives in one disciplined flow instead of being reinvented each time. Read the workflow →
Self-review before the human gets paged
Every new PR runs an auto-review that catches naming, obvious bugs, missing tests, and convention drift. Findings post as draft comments the author can accept or dismiss before tagging a human reviewer. The human gets to focus on intent and architecture, not whitespace.
Careful-review pass before opening the PR
Before opening a PR, /engineering:careful-review re-reads everything the session touched with fresh eyes (bugs, stale comments, dead code, weak tests) and fixes what it finds. Optionally, /engineering:codex-review runs an OpenAI Codex pass against main for an independent second opinion on P0/P1 issues. Then /engineering:ship-pr commits, pushes, and opens the PR with a proper summary and test plan. NEVER skips hooks.
Disciplined diagnosis for the bugs that hide
The bug that resists casual debugging needs a feedback loop, not more staring. /engineering:diagnose imposes the discipline: construct a fast deterministic pass/fail signal first (failing test, curl, replay, bisection harness), then bisect, hypothesise, instrument, fix, regression-test. The first phase is the entire skill; once the signal exists, the cause falls out. Read the discipline →
TDD that doesn't write crap tests
/engineering:tdd insists on vertical slices: one test → one implementation → repeat. No writing all the tests upfront ("horizontal slicing"), because that produces tests that verify shape, not behavior. Tests go through public interfaces, not internals. The bundled tdd/mocking.md, tests.md, and deep-modules.md reference files travel with the skill. Pair with /engineering:find-missing-tests for a coverage-gap survey first.
Deploy checklist that actually blocks
Your CD pipeline POSTs to the routine's /fire endpoint before the prod step. The routine runs /engineering:deploy-checklist: migrations reviewed, tests green, feature flag set, runbook linked, on-call notified. Anything missing returns a non-zero result and the pipeline halts with a Slack message explaining what's left.
Deploy checklist that actually blocks
Your CD pipeline POSTs to the routine's /fire endpoint before the prod step. The routine runs /engineering:deploy-checklist: migrations reviewed, tests green, feature flag set, runbook linked, on-call notified. Anything missing returns a non-zero result and the pipeline halts with a Slack message explaining what's left.
Incident triage that pulls the right context first
When you ack a page, run /engineering:incident-response <incident-id>. Three sub-agents fan out in parallel: recent deploys, error-rate spikes, related alerts. You get a unified timeline in 30 seconds instead of clicking through five tabs.
Data
Validated SQL against your real warehouse. Stakeholder-ready charts. Weekly metric anomalies surfaced for you.
Weekly metric anomaly digest
Monday morning, a scheduled job runs /data:statistical-analysis against your top KPIs (signup, activation, revenue, retention) to surface week-over-week movement that exceeds normal variance. Anything that trips writes a short note to #data-weekly with the slice, the query, and a candidate explanation.
Self-serve queries that don't lie
Anyone on the team can run /data:write-query "revenue by region for Q3, excluding refunds" and get a working query against the real schema, with the unit assumptions made explicit. Plan mode lets you sanity-check before running.
Product
Interview transcripts synthesized in one pass. Competitive scans on a cadence. Spec drafts that match house style.
User-interview synthesis at the end of every research sprint
Point /product-management:synthesize-research at a folder of transcripts. One sub-agent extracts themes per transcript; a coordinator agent rolls them up into a synthesis doc with quote evidence and frequency counts. Reclaim a day of your week.
Weekly competitive scan, ready for the team channel
A scheduled run of /product-management:competitive-brief watches the competitive set you care about, summarizes what shipped or changed, and posts a digest to #product-watch. No more "did anyone see the X launch?"
Finance
Month-end as a workflow. Variance you have to explain, surfaced before the CFO asks. Vendor contract triage that respects the playbook.
Month-end runbook, enforced
The month-end runbook fires automatically: /finance:close-management orchestrates the steps, /finance:reconciliation ties the ledger out, /finance:variance-analysis pulls variance vs plan, /finance:journal-entry-prep drafts the entries. Plan mode lets the controller approve each entry before posting. Half a day saved every month.
Variance you'll have to explain, surfaced before the CFO asks
Every nightly close, the variance skill compares actuals vs plan, flags anything > 5% deviation, drafts the why, and DMs the CFO's prep channel by 7am. The morning standup starts with answers, not surprises.
Vendor contract triage with your playbook in the loop
Any new vendor contract dropped in the legal folder triggers a playbook-aware review: /legal:review-contract walks payment terms, auto-renewal language, liability caps, and data clauses; /legal:vendor-check and /operations:vendor-review add the procurement-side perspective. Findings post as a checklist. Legal + Finance both see what to fight for before the call.
Legal
NDA triage in minutes. Clause-by-clause playbook checks. Drafting that matches your standard language.
NDA triage queue
Inbound NDAs hit /legal:triage-nda first, which classifies the agreement and decides whether /legal:review-contract can auto-redline or whether the deal is non-standard enough to need a human. Non-standard hits get flagged with the specific clause that needs human review. The legal queue drains 5× faster.
Operations
Weekly ops review automated. Runbook adherence enforced. Cross-team status that's never stale.
Weekly cross-team status digest
One sub-agent per department runs /operations:status-report against Linear, Slack, and Notion; the coordinator rolls everything up into a single cross-team status doc by Friday afternoon. Monday standup walks in with the picture already in everyone's head.
People
Recruiting pipeline status. Interview prep that matches the role. Onboarding day-one without the checklist scramble.
Weekly recruiting pipeline digest
/human-resources:recruiting-pipeline pulls the funnel from the ATS, flags candidates aging in a stage, and notes interviews scheduled for the week. The digest lands in #hiring-weekly before Monday standup so hiring managers walk in knowing exactly where every open req stands.
Interview prep tailored to the role and the candidate
When an interview lands on the calendar, /human-resources:interview-prep pulls the JD, candidate resume, and prior round feedback. It drops a prep brief into the event description: focus areas, suggested probing questions, and what to avoid re-asking. Panel walks in calibrated.
Customer Support
Ticket triage in real time. Macro suggestions that match your voice. Escalation routing that respects who's on-call.
Real-time ticket triage with macro suggestion
New ticket fires /customer-support:ticket-triage. It classifies urgency, pulls account context from HubSpot, and lines up /customer-support:draft-response with the right tone for the segment. If the severity bar is high, /customer-support:customer-escalation pings the on-call engineer in Slack. Agent edits the draft, sends. First-response time drops; quality stays high.
Productivity
Cross-app task management. End-of-day update that pulls from everywhere. Project memory that survives context switches.
Daily update that pulls itself together
End of the day, /productivity:update assembles what you shipped (commits, tickets closed, Slack threads contributed to) and drafts a one-paragraph wrap; /productivity:task-management rolls open work into tomorrow's queue. Plan mode lets you edit before either posts to your team's standup channel. Async standups stop being a chore.
Project memory that follows you across sessions
/productivity:start re-hydrates yesterday's context from /productivity:memory-management: open threads, the decisions you made, the things you said you'd come back to. No more spending the first 20 minutes of every session reading old messages.
Design
For frontend code design, Pace defers to impeccable. Use Pace's /design plugin for the UX-research / design-system layer; reach for impeccable when you're writing the UI code itself.
What impeccable is
Impeccable is a separate Apache 2.0 skill kit focused entirely on frontend craft for AI coding agents. One router (/impeccable) with 23 sub-commands across typography, color, layout, motion, accessibility, UX writing, design systems, and dev handoff. It runs in every harness Pace runs in (Claude Code, Cursor, Gemini CLI, Codex CLI, and 8 more).
Pace's site, build pipeline, and multi-harness install machinery were forked from impeccable; the two projects share a maintainer ethos but ship independently.
Per-PR design audit before review
Every PR touching CSS / JSX / Vue templates runs /impeccable audit against the changed files. Findings (contrast issues, anti-pattern hits from impeccable's 50+ detection rules, layout regressions) post as inline comments. Reviewers see the design diff alongside the code diff.
Polish pass on a feature branch
Before merging, run /impeccable polish on the branch. The router orchestrates typography, color, layout, and motion sub-commands; plan mode shows the diff before anything ships. The polish pass is what turns a working feature into one that feels intentional.
Cross-tool design system sync
Pace's /design plugin manages the system at the spec level (Figma tokens, contribution model, dev handoff). Impeccable's /colorize and /typeset apply those tokens in code with full understanding of how they look on screen. Use both: the design plugin sets the rules, impeccable enforces them in the codebase.
When to use Pace's /design plugin vs impeccable
Different layers of the same craft:
- Pace
/designcovers UX research, design system management, accessibility audits at the policy level, dev handoff, and research synthesis. Use it for the work a design team does. - Impeccable covers the visual + interaction craft of writing the UI code itself: typography choices, color tokens, spatial rhythm, motion design, responsive layout. Use it for the work a designer would obsess over when looking at the rendered pixels.
Most teams want both. Install impeccable alongside Pace; the Pace CLI (npx pace skills install) offers it during setup.