glee-fully-repo-standardizer

Scaffold and standardize any Glee-fully child repository against the canonical structure defined in OKHP3/Glee-fullyTools-FoundRy.

When to Activate

Use this skill when the user is working inside (or about to set up) a glee-fully-gpt* child repository and wants to:

• Create the canonical folder/file structure from scratch
• Audit an existing repo for missing folders or files
• Pre-populate files with Glee-fully brand payload (tone, persona, governance)
• Prepare a repo to receive content migrated from ChatGPT threads or Notion

Core Concepts

Every Glee-fully child repo belongs to one of three tiers:

| Tier | Pattern | Emoji | Role | |------|---------|-------|------| | Toolbox | glee-fully-gpt00-* | 🧰 | Trunk — suite concierge, routing only | | Tool | glee-fully-gpt01-* through glee-fully-gpt07-* | 🪚 | Branch — category hub, routing only | | Tool-ette | glee-fully-gpt01a-* through glee-fully-gpt07g-* | 🔩 | Twig — specialized task executor |

Tier determines the folder structure and default tone overlay.

Step 1 — Confirm Project Root

Verify you are at the root of the target child repo:

bash
ls manifest.yaml AGENTS.md README.md 2>/dev/null || echo "Starting fresh"
pwd

Step 2 — Gather Entity Details

Before running the scaffold, collect:

| Field | Example | Source | |-------|---------|--------| | --tier | toolette | Repo name pattern (01a = toolette, 01 = tool, 00 = toolbox) | | --name | "Resume Builder" | Inventory file or GPT Builder | | --id | 01a | Repo name suffix | | --parent | "Discovered Careers" | Parent repo README or inventory | | --parent-url | https://chatgpt.com/g/g-... | Inventory or ChatGPT | | --chatgpt-url | https://chatgpt.com/g/g-... | Inventory or ChatGPT | | --tone | GleeLite | Tier default or explicit override |

Tier defaults if --tone is omitted:

• toolbox → BledsGLEE
• tool → GleeRich
• toolette → GleeLite

Auto-detection: If manifest.yaml already exists in the repo, the script reads name, type, and chatgpt_url from it automatically.

Step 3 — Locate the Script

bash
# Preferred: canonical project-level copy
ls scripts/glee-fully-scaffold.py 2>/dev/null && SCRIPT=scripts/glee-fully-scaffold.py

# Fallback: bundled inside the skill package
SCRIPT=$(python3 -c "
import os, sys
skill_dir = [d for d in [
  '.agents/skills/glee-fully-repo-standardizer/scripts/scaffold.py',
] if os.path.exists(d)]
print(skill_dir[0] if skill_dir else 'NOT FOUND')
")
echo "Script: $SCRIPT"

Step 4 — Run the Scaffold

Full scaffold (new repo):

bash
python3 $SCRIPT \
  --tier toolette \
  --name "Resume Builder" \
  --id 01a \
  --parent "Discovered Careers" \
  --parent-url "https://chatgpt.com/g/g-68578aa0c7d48191b02d8078fd26b9e2-discovered-careers-by-glee-fully" \
  --chatgpt-url "https://chatgpt.com/g/g-6855e58bf8d48191bf27795f6d5ec23c-resume-builder-by-glee-fully"

Preview only (dry run):

bash
python3 $SCRIPT --tier toolette --name "Resume Builder" --id 01a --dry-run

Audit existing repo (show missing files/folders):

bash
python3 $SCRIPT --audit

Overwrite specific files (use sparingly):

bash
python3 $SCRIPT --tier toolette --name "Resume Builder" --id 01a --overwrite

Step 5 — Report Results

After running, tell the user:

• How many folders were created
• How many files were written (new) vs skipped (already existed)
• Which files contain [TODO] placeholders that need filling
• What the recommended next action is (fill gpt/instructions.md first)
• That assets/glee-fully-brand.json was written with shared brand payload

that all files in this repo reference for tone/persona consistency

Folder Structure by Tier

Tool-ette (most common — 42 repos)

AGENTS.md                    ← AI agent navigation (Glee-fully branded)
README.md                    ← Entity overview
CHANGELOG.md                 ← Version history
LICENSE.md                   ← Proprietary license
manifest.yaml                ← Repo metadata + Glee-fully fields

gpt/                         ← The Custom GPT artifact
  instructions.md            ← Full instruction payload (~8,000 chars)
  description.md             ← GPT Builder description (300 chars max)
  starters.md                ← Conversation starters (4 starters)
  knowledge/                 ← Knowledge files attached to the GPT
    .gitkeep

pulsebook/                   ← Quality evaluation (required before 1.0)
  pulsebook-v1-7.md          ← Filled PulseBook (run at PROMPT05)

docs/                        ← Human-readable documentation
  overview.md                ← Plain-language summary
  functions.md               ← All functions/capabilities documented

canon/                       ← Canon compliance
  registry-entry.md          ← !CLAUSE declaration for this entity

origin/                      ← Source migration content (read-only)
  chatgpt-exports/           ← Raw ChatGPT thread extracts
    .gitkeep
  notion-exports/            ← Notion page exports
    .gitkeep

assets/                      ← Shared brand payload + visual identity
  glee-fully-brand.json      ← Canonical tone/persona/vernacular data
  icon.png                   ← Retro butterfly icon (add manually)

archive/                     ← Retired or superseded content
  .gitkeep

Tool (7 repos — routing GPTs, no task execution)

Same structure, except:

• gpt/knowledge/ is omitted (Tools route, they do not hold knowledge files)
• gpt/instructions.md uses the router/branch instruction pattern

Toolbox (1 repo — trunk)

Same as Tool structure, with gpt/instructions.md using the trunk/concierge pattern.

Available Commands

bash
# Full scaffold with all options
python3 $SCRIPT --tier toolette|tool|toolbox --name NAME --id ID \
  [--parent NAME] [--parent-url URL] [--chatgpt-url URL] \
  [--tone OVERLAY] [--dry-run] [--overwrite] [--quiet]

# Audit only — show missing structure, do not write
python3 $SCRIPT --audit

# JSON report of what would be created
python3 $SCRIPT --tier toolette --name NAME --id ID --json

Gotchas

• Run from the repo root. The script writes relative to cwd. Running

from inside a subfolder will create structure in the wrong place.

• Existing files are skipped by default. Pass --overwrite only when you

want to reset a stub. Content you have already added will be lost.

• assets/glee-fully-brand.json is always written (or updated). It is

the shared payload source. All other template files reference its content by comment — they do not import it at runtime (that is the GPT Builder's job).

• origin/ is read-only by convention. Drop ChatGPT exports and Notion

exports there. Never edit files in origin/ — always copy content out to gpt/ or docs/ for refinement.

• gpt/instructions.md is the most important file. It becomes the actual

ChatGPT Builder instruction payload. Fill it first after scaffolding.

• pulsebook/pulsebook-v1-7.md gates 1.0. A repo is not 1.0 until the

PulseBook is filled and manifest.yaml has pme_ready: true.

After Scaffolding — Recommended Fill Order

1. gpt/instructions.md — paste and refine the instruction payload
2. gpt/description.md — 300-char GPT Builder description
3. gpt/starters.md — 4 conversation starters
4. docs/overview.md — plain-language explanation
5. docs/functions.md — list all functions with descriptions
6. canon/registry-entry.md — fill the !CLAUSE block
7. pulsebook/pulsebook-v1-7.md — run evaluation, fill results
8. Update manifest.yaml — set lifecycle_status: active and pme_ready: true

References

• assets/glee-fully-brand.json — canonical brand/tone/persona payload
• OKHP3/Glee-fullyTools-FoundRy/vernacular/ — full vernacular reference
• OKHP3/Glee-fullyTools-FoundRy/evaluation/gpt-pulsebook-evaluation-v1-7.md — PulseBook
• OKHP3/Glee-fullyTools-FoundRy/governance/glee-fully_project_governance_v3-0-1.md — Governance Directive

About

Built for the Glee-fully Personalizable Tools™ ecosystem. OverKill Hill P³ — overkillhill.com Parent FoundRy: OKHP3/Glee-fullyTools-FoundRy