Skip to content

georgeykalangi/calhacks-kit / skills/hack-stack

hack-stack

Verified

Use after the MVP is scoped, or when a hackathon team asks which framework, API, repo, or Claude skill to use. Recommends a small stack from a curated catalog of open-source repos based on the MVP, the team's skills, the hours left, and the sponsor tracks they target. Writes the Stack and architecture sketch into HACK.md.

Hack Stack

Pick the fewest tools that get the one journey working. Every new dependency costs setup time the team does not have.

No file access?

If you are a pasted prompt, ask the team to paste their HACK.md. Use the catalog at https://github.com/georgeykalangi/calhacks-kit/blob/main/skills/hack-stack/catalog.yaml if you can browse; otherwise recommend from what you know and mark each pick "not from catalog, verify".

Steps

1. Read the plan

Read HACK.md: the MVP journey, team strengths, hours remaining, tracks targeted. If there is no HACK.md, tell the team to run /hack-scope first, or ask for those four facts directly.

2. Read the catalog

Read catalog.yaml next to this file. Each entry has use_when, setup_minutes, needs_api_key, license, and demo_value. Treat setup times as rough estimates.

3. Recommend 3-5 picks

Rules:

  • Prefer what the team already knows. A familiar framework beats a better unfamiliar one.
  • Total setup time should stay under 10% of the hours remaining.
  • Prefer tools with free tiers or no keys. Flag anything needing a paid key, a waitlist, or approval.
  • Pick at most two agent-workflow skills. More skills means more context and slower sessions.
  • Flag AGPL licenses if the team mentions continuing the project as a company.
  • If a sponsor track requires a specific API, that API is part of the stack, not optional.
  • You may suggest tools outside the catalog. Mark them "not in catalog, verify before relying on it".

For each pick give: what it does for this MVP, setup minutes, key needed, and the install command only if you are sure of it. Otherwise point to the repo README.

4. Name the riskiest part

Pick the one thing most likely to sink the demo. It can be an integration (an API they have never used, real-time audio, hardware, auth) or an assumption about results (will the model's output be consistent enough to match, rank, or classify on?). Tell the team to spike it for 30 minutes with real inputs before building around it. If the spike fails, change the approach now, not at hour 30.

5. Set up secrets safely

Before any key is used, make sure the repo has:

  • .env listed in .gitignore
  • .env.example with variable names and no values Hackathon repos go public for judging. A key committed at hour 3 is leaked at hour 36.

6. Write HACK.md

Fill the Stack section: picks with one-line reasons, the riskiest integration, and a simple architecture sketch (ASCII or Mermaid) of user -> frontend -> backend -> model/API/data. /hack-ship and /hack-pitch reuse this sketch. Add a Status log line.

Finish with the stack in one line and "Next: spike , then build the journey".

reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 468kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 321kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 321kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 321kUpdated Claude CodeCodexVerifiedInstall