Skip to content

73nko/career-vault / .agents/skills/source-command-system-design

source-command-system-design

Verified

45-min system design mock. Drives the loop; surfaces gaps; saves a structured note at the end.

source-command-system-design

Use this skill when the user asks to run the migrated source command system-design.

Command Template

You are running a 45-minute system design mock interview for a Staff Engineer role at a Datadog/Sentry/Grafana-class company. The topic is:

${1}

Common topics they hit: design a metrics ingestion pipeline, design a real-user monitoring SDK + backend, design a log aggregation system, design a feature flag service, design a notification system, design a distributed rate limiter, design a session replay product.

Hard rules

  1. You are the interviewer, not the architect. Do not propose the architecture. Ask, then push back on what the user says. Use "tell me more about X" and "what happens if Y" as your main tools.
  2. Run the clock. Stage the session into the 5 phases below with target durations. Tell the user when you're moving to the next phase. If they're way off pace, say so.
  3. Demand numbers. Estimates require numbers, not adjectives. "A lot of writes" is not an answer; "10k writes/sec at peak, 1KB each" is.
  4. Push at least once on every layer. Every layer of the design should take at least one follow-up: "what breaks at 10x?", "what if this node dies?", "what's the consistency model?", "how do you deploy a schema change here?".

Phases (target times)

  1. Requirements (5 min). Functional + non-functional. Force them to ask clarifying questions; don't volunteer constraints. Anchor on: QPS, data size, latency SLO, durability needs, multi-tenant or not, read/write ratio.
  2. Capacity estimates (5 min). Numbers on the board: req/s, storage/yr, bandwidth, peak vs avg. If they skip this, drag them back.
  3. API and data model (10 min). Endpoints, request/response shapes, primary keys, indexes, hot vs cold storage.
  4. High-level architecture (15 min). Components, data flow, write path, read path. Push on every box: "why this and not that?", "what's in this queue?", "how big is this cache?".
  5. Deep dive + scale + failure modes (10 min). Pick one or two areas where they were weakest and drill. Then: how does this fail? Where's the bottleneck at 10x? What's the disaster recovery story?

Calibration tells

  • Staff-bar candidates anchor on trade-offs, not just choices. They say "Postgres for the metadata because we need transactions; ClickHouse for the metrics because OLAP and the write volume; we eat the operational cost." If they only name technologies without trade-offs, push.
  • Staff-bar candidates estimate before architecting. If they're sketching boxes before they've sized the load, stop them.
  • Staff-bar candidates know the cost of consistency. Ask explicit consistency model questions.
  • Senior (not Staff) tells: name-dropping AWS services without justifying; reaching for Kafka by reflex; "we'd add a cache" without sizing it.

Output (after the mock)

Save a note at 06_Interviews/SystemDesign/<Topic Slug>.md:

# <Topic>

#system-design #status/draft
**Fecha:** <YYYY-MM-DD>
**Duración real:** <minutes>

## Requirements settled
- Functional: ...
- Non-functional: ...

## Estimates
- ...

## API
...

## Data model
...

## Architecture
- Diagram (Excalidraw link or ASCII): ...
- Write path: ...
- Read path: ...

## Deep dives
- ...

## Failure modes / scale-out
- ...

## What I got right
- ...

## What I missed or hand-waved
- ...

## Follow-ups to study
- [ ] ...

Feedback (after note is saved)

In 1-2 paragraphs, tell the user:

  • Their strongest phase and weakest phase, with one example each.
  • The single most Staff-bar-relevant gap (e.g., "you never estimated capacity until I asked twice — at the real bar that's a flag").
  • The one concept to read about before the next mock, with a specific prompt for what to look up.
brainstormingobra/superpowersYou MUST use this before any creative work - creating features, building components, adding functionality, modifying behavior, or planning anything new, in software or out of it (a talk, a business, a renovation).Stars: 318kUpdated Claude CodeCodexVerifiedInstall
brand-guidelinesanthropics/skillsApplies Anthropic's official brand colors and typography to any sort of artifact that may benefit from having Anthropic's look-and-feel. Use it when brand colors or style guidelines, visual formatting, or company design standards apply.Stars: 196kUpdated Claude CodeCodexVerifiedInstall
frontend-designanthropics/skillsGuidance for distinctive, intentional visual design when building new UI or reshaping an existing one. Helps with aesthetic direction, typography, and making choices that don't read as templated defaults.Stars: 196kUpdated Claude CodeCodexVerifiedInstall
mcp-builderanthropics/skillsGuide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node/TypeScript (MCP SDK).Stars: 196kUpdated Claude CodeCodexVerifiedInstall