ai-asset-pricing: Skill for Claude Code

Create New Research Project

Four phases: Validate → Gather Context → Create Scaffold → Setup Paper.

Examples

• /new-project momentum_replication -- create project with name only
• /new-project vol_surface "Implied volatility surface dynamics" -- name + description
• /new-project bond_liquidity -- another example

Phase 1: Validate Input

Parse $ARGUMENTS:

• First token = project name
• Remaining tokens (if any) = optional description text

If $ARGUMENTS is empty, use AskUserQuestion to ask:

"What should the project be called? Use lowercase with underscores (e.g., momentum_replication, vol_surface, bond_liquidity)."

Validation rules:

• Must match regex: ^[a-z][a-z0-9_]{1,49}$ (lowercase, underscores only, starts with letter, 2-50 chars)
• Reject Windows-reserved names: con, prn, aux, nul, com1-com9, lpt1-lpt9
• If invalid, explain the rules, show examples, and re-ask

Collision check:

• If projects/<name>/ already exists, ask the user: "Project <name> already exists. Do you want to open it, or choose a different name?"
• Do NOT overwrite existing projects

Phase 2: Gather Context

If no description was provided in $ARGUMENTS, use a single AskUserQuestion with these questions:

1. "Brief description of the project (1-2 sentences):"
2. "Which WRDS databases will this project use? (e.g., CRSP, OptionMetrics, Compustat, TAQ, Bonds, JKP)"
3. "Any initial methodology notes or reminders for this project?"

If a description WAS provided in $ARGUMENTS, still ask questions 2 and 3.

Store the answers for template filling.

Phase 3: Create Scaffold

3a. Create directories

Run a single Bash command:

mkdir -p projects/<name>/{latex,code,scripts/tests,results/{figures,tables},literature,guidance,_misc}

3b. Write project README.md

Use the Write tool to create projects/<name>/README.md:

# {Project Name in Title Case}

{User's description}

## Status

- [ ] Data sourced
- [ ] Exploratory analysis complete
- [ ] Main results produced
- [ ] Write-up drafted
- [ ] Write-up finalized

## Data Dependencies

This project uses datasets from the global `data/` folder:

| Dataset | Description | Status |
|---------|-------------|--------|
| *(to be filled as data is fetched)* | | |

## Structure

| Folder | Purpose |
|--------|---------|
| `latex/` | LaTeX writeup |
| `code/` | Production code (clean, reusable) |
| `scripts/tests/` | Exploratory investigations (each in own subfolder with `output/`) |
| `results/` | Publication-ready figures and tables |
| `literature/` | Reference papers |
| `guidance/` | Methodology notes |
| `_misc/` | Catch-all |

## Notes

- Created: {today's date, YYYY-MM-DD}
- WRDS databases: {databases}