HACD Incubator AI Issuance Skill

Bitcoin proved PoW for money. HACD brings PoW to assets.

An AI-assisted issuance workflow for the HACD Labs Incubator Cohort 2 campaign.
Turn a raw project idea into a complete, validated HACD Stack Token launch package — in one AI session, no coding required.

Apply: https://hacd.it/incubator
Launchpad: https://hacd.it/launchpad
Campaign rules + $500 reward pool: see CAMPAIGN.md

What this is

This repo is an AI skill — a set of structured prompts and templates you load into Claude or ChatGPT. Once loaded, the AI becomes a HACD Stack issuance expert that knows the full Hacash ecosystem, the correct terminology, the supply math rules, and every safety check required before a project can go live on the Launchpad.

It generates 8 documents per project. It validates the math. It self-reviews for unsafe claims. It scores submissions so HACD Labs can rank them objectively.

You give it:  a project idea + 9 Google Form answers
It gives you: 8 complete launch documents, validated and review-ready
Time:         under 10 minutes

Live example projects

Two complete working packages are included in this repo. Both pass the validator with no errors.

These are not mockups. Each folder contains all 8 required documents. Run the validator on either one:

python3 scripts/validate_launch_spec.py examples/example_community_token/launch_spec.json
python3 scripts/validate_launch_spec.py hacashbuilders/launch_spec.json

Incubator Cohort 2 — $500 reward pool

Campaign open: 20 June 2026
Campaign close: 27 June 2026
Winners announced: 29 June 2026

Top 3 projects that launch on hacd.it/launchpad win from the prize pool:

See CAMPAIGN.md for full rules, judging criteria, and submission format.

Fastest start — one prompt (BUILD.md)

New here? Don't read 13 prompts. Do this:

1. Load the skill. Open claude.ai or chatgpt.com and paste SKILL.md as your
   first message. (In Claude Code / claude.ai with skills, just invoke the
   hacd-issuance skill — no paste.)
2. Scaffold the folder so you never start from a blank page and the supply math
   is correct before you write a word:

   python3 scripts/new_project.py --name "Your Project" --ticker TKR \
       --type FT --lots 100 --units 10000 --cost 50
   

3. Run the one-shot prompt in BUILD.md: fill the project block, paste,
   and the AI returns all 8 documents, self-reviewed, in a single turn.
4. Validate:

   python3 scripts/validate_launch_spec.py project_tkr/launch_spec.json --strict
   

That's the whole loop. The step-by-step version below still works if you prefer it.

Step-by-step start — 10 minutes

Full guide in QUICKSTART.md. Short version:

Step 1 — Load the skill
Open claude.ai or chatgpt.com. Paste the contents of SKILL.md as your first message.

Step 2 — Expand your application
Paste prompts/google_form_to_intake.md + your 9 Google Form answers. The AI expands them into a full 40-field intake form automatically.

Step 3 — Generate all 8 documents
Tell the AI: "Generate all 8 documents using this intake form."

Step 4 — Validate

python3 scripts/validate_launch_spec.py your_project/launch_spec.json

Step 5 — Self-review
Paste prompts/roast_mode.md + all 8 documents. Fix every issue it finds.

Step 6 — Submit
Send your complete package to HACD Labs via hacd.it/incubator.

Use with Claude Code (ccr)

If you have Claude Code Router installed:

ccr start
ccr code "Load the system prompt from prompts/acp_agent_system_prompt.md and help me design a Stack Token for [your project name] with [N] lots"

The ACP agent system prompt at prompts/acp_agent_system_prompt.md contains full Hacash ecosystem knowledge including live reference data from Carat Protocol, HAC market links, wallet and explorer links, and real stack cost benchmarks.

Prompts reference — 13 prompts total

Key reference docs

Validator — now does more than math

scripts/validate_launch_spec.py checks structure, enums, the supply equation, the phase-lot
sum, cross-document consistency (numbers in the .md files must match the spec), and runs a
negation-aware copy-safety linter that flags unsafe promo language without false-flagging
risk disclosures.

python3 scripts/validate_launch_spec.py your_project/launch_spec.json            # draft check
python3 scripts/validate_launch_spec.py your_project/launch_spec.json --strict   # final / CI check
python3 scripts/validate_launch_spec.py your_project/launch_spec.json --no-docs  # JSON only

Every launch_spec.json in the repo is validated automatically by GitHub Actions on push and PR
(.github/workflows/validate.yml).

Repo structure

hacd-incubator-ai-issuance-skill/
├── README.md
├── ECOSYSTEM.md             ← single source of truth: facts, links, benchmarks, campaign
├── CAMPAIGN.md              ← campaign rules, reward pool, timeline
├── BUILD.md                 ← one-shot mega-prompt: idea → 8 docs, self-reviewed
├── QUICKSTART.md            ← 10-minute step-by-step document-generation guide
├── LAUNCH_WALKTHROUGH.md    ← on-chain guide: wallet → HAC → HACD → Stack → verify
├── SKILL.md                 ← core AI operating instructions (invocable skill w/ frontmatter)
├── LICENSE                  ← MIT
├── prompts/                 ← 13 prompts incl. web_research.md (see table above)
├── templates/               ← blank document templates
├── schema/
│   └── launch_spec.schema.json   ← JSON Schema for launch_spec.json
├── examples/
│   ├── example_community_token/   ← StackFire (SFR) — FT reference
│   └── example_stack_spec.json
├── hacashbuilders/          ← HacashBuilders (BUILD) — FT+NFT reference
├── lovable-prompt-builder/  ← companion skill: idea → executive Lovable demo prompt
│   ├── SKILL.md
│   ├── prompts/             ← idea_board, pow_fit_check, sprint_plan
│   ├── templates/           ← lovable_prompt_template
│   └── examples/proofpress/ ← worked example (PRSS)
├── hacashbuilders/          ← HacashBuilders (BUILD) — HYBRID reference
├── scripts/
│   ├── new_project.py             ← scaffold project_<ticker>/ + prefilled launch_spec.json
│   └── validate_launch_spec.py    ← structure + math + cross-doc + copy-safety linter
└── .github/
    └── workflows/validate.yml     ← CI: validates every spec on push/PR

Companion skill — Lovable Prompt Builder

The issuance skill produces your launch documents. The companion
lovable-prompt-builder/ skill produces your demo
app: it turns the same idea into an executive-grade
Lovable build prompt you paste into Lovable (or hand to
Claude) to ship a working demo a reviewer can watch run in under a minute.

Project idea ─▶ lovable-prompt-builder ─▶ executive Lovable prompt ─▶ working demo app
                                                                   └─▶ pairs with the 8-doc issuance package

It ships the full Cohort 2 idea board (20 vetted Stack Asset concepts), a
PoW-fit check, a sprint plan keyed to CAMPAIGN.md, the fill-in prompt scaffold,
and a complete worked example (ProofPress). Load
lovable-prompt-builder/SKILL.md the same way
you load the root SKILL.md, then give it your idea or an idea-board ticker.

Hacash ecosystem links

What the Skill does not do

• Does not ask for private keys, seed phrases, or wallet passwords
• Does not sign transactions
• Does not guarantee Launchpad approval
• Does not provide legal, financial, or investment advice
• Does not promise price appreciation, yield, or profits

License

MIT — see LICENSE. Fork it, adapt it, build with it.