trilogy-group/codex-second-brain-starter-kit — Summary

trilogy-group/codex-second-brain-starter-kit is a GPT-5.5-powered "second brain" workflow kit for OpenAI Codex, designed to build Obsidian knowledge vaults from multi-source product intelligence. The framework ships 3 skills (product-intelligence-factory, obsidian-intelligence-system, product-engineering-ops), 6 prompt files (bootstrap, ingest, engineering layer, bug workflow, automations, readiness review), bootstrap scripts, and manifest/template YAML. It uses gpt-5.5 with explicit reasoning effort routing: medium for high-fanout reducers, high for synthesis tasks. The central artifact is a product_ontology.json + Obsidian vault combining PARA methodology, CODE workflow, progressive summarization, intermediate packets, and evidence cards from 10+ source types (repos, DOCX/PDF, URLs, support articles, wiki pages). 0 GitHub stars, MIT license, maintained by Trilogy Group's engineering team.

differs_from_seeds: No direct seed equivalent — this is a knowledge management system targeting OpenAI Codex (not Claude Code), built for enterprise product intelligence workflows. Closest seeds: ccmemory (external knowledge store) and claude-conductor (documentation-as-first-class-artifact), but codex-second-brain is orders of magnitude more complex. The BASB (Building a Second Brain) methodology integration (CODE/PARA/progressive summarization) applied to software product intelligence is unique in the entire research corpus. Tyler provider gateway routing for enterprise deployment is an enterprise feature not present in any seed.

trilogy-group/codex-second-brain-starter-kit — Overview

Origin

Published by Trilogy Group (enterprise software company). 0 GitHub stars. MIT license. Python language. Last pushed 2026-05-12. 1 contributor (Colin Guilfoyle, Sergio Wermuth listed as maintainers).

Philosophy

From the README:

"This is a generic Codex setup for testing the second-brain workflow on any codebase."

"This package is intentionally generic. It does not assume any specific repository, product, or support system."

The framework applies Tiago Forte's BASB (Building a Second Brain) methodology to software product intelligence:

• CODE: Capture, Organize, Distill, Express
• PARA: Projects, Areas, Resources, Archives
• Progressive Summarization: Layer-by-layer knowledge compression
• Intermediate Packets: Reusable knowledge units

Intelligence Model

From the README:

"Full-fidelity LLM synthesis defaults to gpt-5.5 with hybrid reasoning: medium for high-fanout source/theme reducers, generation shards, and business-value entity synthesis, and high for capability reducers, ontology reducers, Product Ontology v2, and semantic cluster synthesis."

The framework uses OpenAI's Responses API (not Chat Completions) for explicit reasoning effort routing.

Enterprise Deployment: Tyler Provider Gateway

TYLER_PROVIDER_GATEWAY_BASE_URL: ...
TYLER_PROVIDER_GATEWAY_TOKEN: ...
TYLER_PROVIDER_GATEWAY_PRODUCT_ID: ...
TYLER_PROVIDER_GATEWAY_JOB_RUN_ID: ...

When Tyler environment variables are set, all OpenAI calls route through Trilogy's provider gateway for durability, retries, and live diagnostics.

Evidence Types Supported

Repositories, repo docs, uploaded files, crawled URLs, support articles, wiki pages, DOCX/PDF/plain documents, code references, generated notes — combined into a unified multi-source evidence-card layer.

trilogy-group/codex-second-brain-starter-kit — Architecture

Distribution

Standalone GitHub repository. Install skills into Codex home directory.

Install Method

# Copy skills to Codex home
cp -r skills/ $CODEX_HOME/skills/

# Install Python dependencies
pip install -r requirements-dev.txt  # optional: tree_sitter, tree_sitter_languages

# Optional GitHub CLI for repo mirror sync
brew install gh

Directory Tree

trilogy-group/codex-second-brain-starter-kit/
├── skills/
│   ├── product-intelligence-factory/
│   │   ├── SKILL.md
│   │   ├── agents/
│   │   │   └── openai.yaml        (OpenAI Responses API config)
│   │   ├── references/
│   │   │   ├── manifest-schema.md
│   │   │   └── automation-pack.md
│   │   └── scripts/
│   ├── obsidian-intelligence-system/
│   │   ├── SKILL.md
│   │   └── ...
│   └── product-engineering-ops/
│       ├── SKILL.md
│       └── ...
├── scripts/
│   ├── init_product_manifest.py    (creates product.yaml)
│   ├── validate_product_manifest.py
│   ├── init_intelligence_profile.py
│   ├── vault_rebuild.py
│   ├── source_index.py
│   └── privacy_sanitizer.py
├── prompts/
│   ├── 01_bootstrap_product.md
│   ├── 02_ingest_and_build_second_brain.md
│   ├── 03_build_engineering_layer.md
│   ├── 04_bug_workflow.md
│   ├── 05_create_automations.md
│   └── 06_review_readiness.md
├── templates/
│   ├── manifest.yaml
│   ├── profile.yaml
│   └── automation templates
├── examples/
│   └── sample manifests
├── INSTALL.md
├── QUICKSTART.md
├── WIZARD.md
├── WORKFLOW.md
└── requirements.txt

Required Runtime

• openai-codex (cloud platform)
• Python 3
• Git
• Obsidian (for vault)
• OPENAI_API_KEY (for gpt-5.5 semantic clustering)
• Optional: tree_sitter, tree_sitter_languages (for AST-backed code intelligence)
• Optional: GitHub CLI (gh) for repo mirror sync

Config

• product.yaml — product manifest (product identity, vault/workspace paths, repos, sources)
• intelligence-profile.yaml — synthesis settings (model, reasoning effort, workers)
• Tyler env vars for enterprise gateway routing

trilogy-group/codex-second-brain-starter-kit — Components

Skills (3)

product-intelligence-factory

• Builds repeatable product-import systems for multiple products
• Creates product manifests via init_product_manifest.py
• Validates manifests, creates capability profiles
• Manages automation packs
• Pairs with obsidian-intelligence-system and product-engineering-ops

obsidian-intelligence-system

• Builds Obsidian vaults from product intelligence
• Implements BASB (Building a Second Brain) methodology
• Handles CODE stages, PARA mapping, progressive summarization
• Manages output conversion and archive hygiene
• Processes evidence cards from multi-source inputs

product-engineering-ops

• Code intelligence inventories: routes, schemas, calls, dependencies, tests
• Ownership, churn, AST/parser backends analysis
• Engineering readiness reporting
• Bug workflow management

Scripts (6)

• init_product_manifest.py — creates product.yaml with product identity, paths, repos, sources
• validate_product_manifest.py — validates manifest schema and path existence
• init_intelligence_profile.py — creates intelligence-profile.yaml with gpt-5.5 settings
• vault_rebuild.py — source-index + privacy-sanitizing vault rebuild
• source_index.py — creates source index for vault content
• privacy_sanitizer.py — sanitizes sensitive content before vault storage

Prompts (6, ready-to-paste into Codex)

• 01_bootstrap_product.md — validates manifest, confirms BASB model, checks workspace
• 02_ingest_and_build_second_brain.md — builds evidence-card layer from all sources
• 03_build_engineering_layer.md — code intelligence inventory (routes, schemas, AST)
• 04_bug_workflow.md — bug workflow with code intelligence context
• 05_create_automations.md — lifecycle automations (weekly reviews, archive records)
• 06_review_readiness.md — engineering readiness report generation

Agents

• skills/product-intelligence-factory/agents/openai.yaml — OpenAI Responses API configuration
  • Model: gpt-5.5
  • reasoning_effort: medium (high-fanout reducers) / high (synthesis)

Templates

• templates/manifest.yaml — product manifest template
• templates/profile.yaml — intelligence profile template with reasoning effort settings

Central Artifact: product_ontology.json

A canonical JSON artifact containing: product purpose, personas, capabilities, workflows, repos, APIs, entities, tests, CI/CD, evidence citations. The ontology is the persistent structured representation of a product's intelligence.

trilogy-group/codex-second-brain-starter-kit — Uniqueness

differs_from_seeds

No direct seed equivalent. The closest seeds are ccmemory (external knowledge store, vector DB) and claude-conductor (documentation-as-artifact), but codex-second-brain exceeds both in scope: it combines Obsidian vault management, multi-source evidence cards, semantic clustering, product ontology generation, and BASB methodology into a single enterprise-scale workflow. The BASB methodology (CODE/PARA/progressive summarization/intermediate packets) applied to software product intelligence is unique in the entire research corpus. Explicit reasoning effort routing (medium vs high per task type) using OpenAI's Responses API (not Chat Completions) is more sophisticated than any seed's model configuration. The Tyler provider gateway for enterprise Codex deployment is an enterprise feature not present in any framework in this analysis.

Positioning

Enterprise product intelligence system for Codex power users at technology companies. Trilogy Group's own engineering teams are the primary users. The 0-star count reflects a newly published internal toolkit made public rather than a community project.

Observable Failure Modes

1. Complex dependencies: Requires Obsidian, Python 3, OpenAI API key, optional tree_sitter, optional GitHub CLI. High setup cost.
2. OpenAI-only: Locked to gpt-5.5 and OpenAI Responses API. Cannot be used with Claude or other providers without significant refactoring.
3. Obsidian requirement: Obsidian is required for the vault UI. Enterprise environments without Obsidian licenses cannot use this framework as-is.
4. Tyler-specific enterprise features: The Tyler provider gateway is Trilogy-internal infrastructure. External users cannot use Tyler routing without Trilogy's platform.
5. No Claude Code support: This is a Codex framework. Claude Code users cannot use it directly.

Comparison with codex-starter-acthouse

Both target OpenAI Codex, but they're at opposite ends of complexity:

• codex-starter-acthouse: 3 files, minimal setup, beginner-oriented
• codex-second-brain-starter: 50+ files, enterprise intelligence system, expert-oriented