# Codex — VIBECODE-THEORY Project

You are an autonomous agent building structural tools for a philosophy paper series about AI, technology dependence, and the singularity. Work independently — read, build, output.

## Tool Mapping

| Instructions say | You use |
|-----------------|---------|
| `Bash` | `shell` |
| `Read` | `read_file` |
| `Write` | `write_file` |
| `Grep` | `shell` with `grep` |
| `Glob` | `shell` with `find` |

## Bootstrap

1. Read `CODEX_TASKS.md` — claim the first OPEN task (edit status to `IN PROGRESS`, write your name in `Claimed by`)
2. Read all source papers (`001*.md` through `008*.md` + `allegorical/`) — these are your input
3. Build the deliverables specified in your task — output goes in `tools/`
4. Mark task `DONE` in `CODEX_TASKS.md`
5. If another task is OPEN, claim and repeat

## Project Structure

```
VIBECODE-THEORY/
├── 001-008*.md          # Paper series (input, read-only)
├── allegorical/         # 8 allegory files (input, read-only)
├── research/            # Gemini research output (don't touch)
├── CODEX_TASKS.md       # Your work queue
├── tools/               # Your output goes here
└── WORKFLOW.md          # Context only
```

## Rules

- Don't modify paper files or anything in `research/`
- Python stdlib preferred. If you need a pip package, install it first.
- Self-contained, re-runnable scripts where applicable

## Project Structure

```
VIBECODE-THEORY/
├── 001-008*.md          # The paper series (your input)
├── allegorical/         # 8 allegory files (your input)
├── research/            # Gemini research output (may be empty or in-progress)
├── CODEX_TASKS.md       # Your work queue
├── RESEARCH_TASKS.md    # Gemini work queue (don't touch)
├── tools/               # Your code output goes here
└── WORKFLOW.md          # How papers get written (context only)
```

## Rules

1. **Read before building.** Parse the actual paper files. Don't assume content.
2. **Don't modify papers.** The `.md` paper files are source material. Read-only.
3. **Don't touch Gemini files.** `RESEARCH_TASKS.md`, `GEMINI.md`, and `research/` are theirs.
4. **Output goes in `tools/`.** All scripts, generated data, and output files go in the `tools/` directory.
5. **Self-contained scripts.** Use Python stdlib or minimal dependencies. The environment has Python 3, jq, and standard Unix tools. Don't assume pip packages are installed — if you need one, install it first.
6. **Claim your task.** Edit `CODEX_TASKS.md` before starting work.