Project Intake - Codex Skills Library

1) Executive Summary

2) Project Type and Domain

Project type: Program / tooling library.
Domains: AI workflows, developer tools, documentation systems, agent operations, repo-native context.
Primary users: Maggie Lerman as operator/maintainer; future Codex users who inspect, copy, adapt, or contribute skill workflows; agents working inside repos that need reusable operating procedures.
Project posture: Public-learning-oriented repository with all rights reserved until an explicit public license is selected.

3) Repository and Live URLs

Repository: https://github.com/maggielerman/codex-skills
Local docs review: http://localhost:3030
Public docs deployment: Unknown in repo metadata reviewed for this intake.

4) Product/User Journey Summary

5) Architecture and Stack Summary

Source model: skills/ contains portable skill folders; skills/SUITE_SKILLS.txt is the curated suite allowlist; skills/SUITE_METADATA.json tracks lifecycle labels; plugins/ backs up selected custom local plugins.
Generated catalogs: scripts/build_catalog.py validates suite drift and regenerates skills/INDEX.md plus skills/manifest.json; scripts/build_docs_site_catalog.py regenerates docs-site/src/lib/catalog.generated.ts.
Docs site: docs-site/ is a Next.js 16, React 19, TypeScript, Tailwind, shadcn/base-ui, and lucide-react application for public browsing.
Governance docs: AGENTS.md, README.md, CONTRIBUTING.md, ROADMAP.md, CHANGELOG.md, and DOCS/ define repo maintenance expectations, project memory, and evidence conventions.
Validation: The normal validation path is catalog drift checking, generated catalog rebuilds, docs-site catalog regeneration, lint/build where relevant, and targeted script checks.

6) Key Workflows and Operational Flows

Skill sync and drift review: Run ./scripts/sync_from_codex_home.sh --check to compare allowlisted repo skills against local Codex/Agents skill homes, then use --prune-stale only when stale repo files should be removed.
Catalog validation: Run python3 scripts/build_catalog.py --check before regeneration to detect missing allowlist entries, stale folder state, or generated catalog drift.
Catalog regeneration: Run python3 scripts/build_catalog.py and python3 scripts/build_docs_site_catalog.py after intended skill/plugin changes so the generated suite and public docs stay in sync.
Plugin backup maintenance: Keep custom plugins self-contained under plugins/, preserve .codex-plugin/plugin.json, refresh .agents/plugins/marketplace.json, and regenerate the docs-site catalog when plugin metadata changes.
Public docs review: Run npm run dev or npm run build in docs-site/ to inspect generated catalog pages before treating the public documentation layer as current.

7) AI Context: Prompts, Chats, Agent Workflows

This repository is itself an AI-agent operating artifact. Skills encode when an assistant should use a workflow, what files and references to read, what scripts to prefer, and how to verify completion. The local AGENTS.md makes the repository's agent contract explicit: treat skills/ as the primary product, avoid importing hidden/system skills, do not silently rewrite imported skills for tone, preserve third-party notices, and keep plugin backups separate from the curated skill suite.

Current intake evidence also shows a maintenance boundary: the repo worktree already contains unrelated in-progress edits, including an unlisted hyphenomenon-chat-intake skill folder. This intake records the repo as-is without resolving that separate catalog drift.

8) Supporting Docs and External Resources

README.md explains the repository purpose, layout, sync model, generated catalog, docs site, and support direction.
AGENTS.md defines operating assumptions, required workflows, scope boundaries, quality bar, and docs/evidence conventions for agents.
scripts/build_catalog.py is the key source of truth for catalog validation and generated suite outputs.
docs-site/src/app/page.tsx and docs-site/src/app/docs/page.tsx define the public browsing surfaces used for screenshots.
plugins/context-layer/.codex-plugin/plugin.json proves the repo also preserves custom plugin metadata, not only standalone skills.

Hydration Source Artifacts

Artifact (relative repo path)	Canonical URL	Type	Why hydrate this into note body
`README.md`	https://github.com/maggielerman/codex-skills/blob/main/README.md	markdown	Establishes the project purpose, repo layout, generated catalog contract, sync commands, docs-site role, and support direction.
`AGENTS.md`	https://github.com/maggielerman/codex-skills/blob/main/AGENTS.md	markdown	Captures the agent operating contract, required skill/plugin workflows, scope boundaries, quality bar, and docs/evidence expectations.
`scripts/build_catalog.py`	https://github.com/maggielerman/codex-skills/blob/main/scripts/build_catalog.py	text	Documents the executable catalog validation/regeneration system that keeps suite metadata, index, and manifest aligned.
`docs-site/README.md`	https://github.com/maggielerman/codex-skills/blob/main/docs-site/README.md	markdown	Explains the public docs-site intent, local development commands, generated catalog relationship, and support model.
`plugins/context-layer/.codex-plugin/plugin.json`	https://github.com/maggielerman/codex-skills/blob/main/plugins/context-layer/.codex-plugin/plugin.json	json	Shows how custom local plugin backups are represented as durable repo artifacts.
`DOCS/intake/screenshots/01-home.png`	https://github.com/maggielerman/codex-skills/blob/main/DOCS/intake/screenshots/01-home.png	image	Preserves visual proof of the docs-site home route captured from the local review server.

9) Screenshot Gallery (with relative links + captions)

Home/Landing: - Public-facing home route for the Codex skills library and workflow notes.
Primary Workflow: - Library docs and generated skill catalog browsing flow.
Key Feature: - Generated skill detail route for the Hyphenomenon project-intake workflow.
Admin/Settings Equivalent: - Generated plugin backup detail route showing larger bundled workflow metadata.

10) Candidate Import Nodes and Relationships

Project supernode: codex-skills-library / Codex Skills Library.
Candidate project record: skills and plugin library for reusable Codex workflow patterns, documentation systems, local plugin backup, and public catalog browsing.
Supporting notes: README overview, agent operating contract, catalog builder, docs-site README, context-layer plugin metadata, and docs-site screenshot proof.
Workflow nodes: skill sync and catalog regeneration workflow; public docs-site catalog publishing workflow.
Data model nodes: skill suite catalog model; custom plugin backup model; docs-site generated catalog model.
Relationships: project relates to AI workflow infrastructure, context-layer work, Hyphenomenon project-intake skill, RPS Etsy operations skills, docs-system scaffolds, and public working-notes/skills evidence.

11) Risks, Gaps, and Unknowns

Public docs deployment URL was not confirmed from the reviewed repo files; screenshots use the local docs server at http://localhost:3030.
The worktree has unrelated in-progress edits and catalog drift: python3 scripts/build_catalog.py --check reports hyphenomenon-chat-intake is present in skills/ but missing from skills/SUITE_SKILLS.txt.
No .docx hydration artifact was identified.
The repository has PDF audit output under output/pdf/, but this local-only intake did not hydrate it because the core repo contract is better represented by source docs and scripts.
License remains intentionally unresolved: the README says all rights remain reserved until a public license is selected.

12) Handoff Checklist for Hyphenomenon Import

Confirm dossier sections complete and source-backed.
Confirm screenshot files exist in DOCS/intake/screenshots.
Confirm screenshot links render from DOCS/intake/hyphenomenon-project-intake.md.
Confirm unknowns are explicit.
Run Hyphenomenon extraction and validation against this dossier.
Run local-only dry-run and apply against the local Hyphenomenon knowledge backbone.
Do not apply to production unless a future operator checkpoint explicitly approves production mutation.

Dataset Preview

Raw CSV row/table content is available in the source artifact.