Profiled Prompt System

Profiled Prompt System is a Codex skill for turning rough user requests into profile-aware AI workflows.

It sits above a basic "question to prompt" workflow and adds:

• user-group routing for researchers, developers, creators, operators, students, and knowledge workers
• research-specific prompt frames and quality gates
• AI Work Receipts for auditable task records
• Prompt Escrow for freezing task contracts before long or risky work
• Artifact Passports for durable outputs such as papers, figures, decks, code releases, and datasets
• Obsidian-friendly workflow indexes

The main use case is research and technical work where AI output should be traceable, reproducible, and easy to continue later.

What It Does

raw request
-> profile lens
-> phase detection
-> prompt pack
-> skill route
-> execution contract
-> receipt / escrow / passport
-> Obsidian index

Example:

python scripts/profile_router.py "read this paper into an Obsidian note and leave a receipt"

Output:

profile: researcher
phase: reading
modules: AI Work Receipt, Obsidian Deal Room
recommended_skills: nature-reader, literature-review, paper-lookup
confidence: high

Repository Layout

Path	Purpose
SKILL.md	Codex skill entrypoint
references/researcher-profile.md	research workflow router
references/research-prompt-templates.md	reusable research prompts
references/research-quality-gates.md	quality gates for literature, experiments, writing, figures, and PPT
references/prompt-system-modules.md	receipt, escrow, passport, regret pack, and workflow ledger modules
scripts/profile_router.py	rough request to profile/phase/modules/skills
scripts/make_ai_work_artifact.py	generate receipts, escrows, passports, regret packs, and deal rooms
scripts/update_obsidian_index.py	generate an Obsidian index from AI work artifacts
examples/	ready-to-read demos

Install As A Codex Skill

Copy this folder into your Codex skills directory:

Copy-Item -Recurse . C:\Users\Admin\.codex\skills\profiled-prompt-system

Validate:

python C:\Users\Admin\.codex\skills\.system\skill-creator\scripts\quick_validate.py C:\Users\Admin\.codex\skills\profiled-prompt-system

CLI Usage

Route A Rough Request

python scripts/profile_router.py "plan a research experiment and create a receipt"

Generate An AI Work Receipt

python scripts/make_ai_work_artifact.py --kind receipt --auto --profile researcher --phase experiment --project . --out .\ai-work --skill profiled-prompt-system --verification "tests passed"

Generate Prompt Escrow

python scripts/make_ai_work_artifact.py --kind escrow --profile researcher --phase writing --project . --out .\ai-work

Generate An Obsidian Index

python scripts/update_obsidian_index.py --root .\ai-work --out ".\AI Workflow Index.md"

Research Workflow

Research phase	Default artifact	Suggested downstream skills
reading	source-grounded reading note	nature-reader, literature-review, paper-lookup
search	literature search contract	paper-lookup, literature-review, nature-academic-search
idea	hypothesis card	brainstorming-research-ideas, creative-thinking-for-research
experiment	experiment contract + run log	codex-execution-loop, domain-specific execution skills
figure	figure contract + artifact passport	nature-figure, academic-plotting
writing	claim-evidence outline	nature-writing, nature-polishing, ml-paper-writing
review	attack surface + response plan	nature-response, innovation-to-proof-plan
presentation	slide story contract	nature-paper2ppt
knowledge-base	Obsidian note with links and tags	local note workflow

Examples

• Research reading workflow
• Experiment receipt workflow
• Developer change receipt workflow
• Obsidian workflow index

Design Principles

• Keep the first prompt executable.
• Use profile adaptation only when it changes the workflow.
• Freeze high-risk tasks before execution.
• Record what AI did, not just what it said.
• Treat research claims as evidence-bearing objects.
• Prefer local files and transparent Markdown over hidden memory.

Status

This is an early but usable skill package. The strongest current path is researcher workflows: reading, experiment planning, writing, artifact tracking, and Obsidian indexing.