Skip to content

Cytognosis Documentation Standards and Lifecycle Framework

Status: Active Date: 2026-05-25 Author: Antigravity Tags: standards, documentation, lifecycle, obsidian-vault

This document defines the technical documentation standards and lifecycle framework for the Cytognosis Foundation. We track two separate, parallel versions of our knowledge assets side-by-side to serve different cognitive styles and engineering needs. This framework ensures our codebase remains transparent, maintainable, and aligned with our scientific mission.


1. The Side-by-Side Dual-Track Strategy

We maintain two distinct representations for every major documentation asset. This approach bridges the gap between rigorous technical specifications and rapid conceptual comprehension.

graph TD
    A["Engineering Decision / Requirement"] --> B["Dual-Track System"]
    B --> C["Track A: Technical Specs (Docs-as-Code)"]
    B --> D["Track B: ADHD-Friendly Visual Specs"]
    C --> E["Git Repository (/design)"]
    D --> F["Obsidian Vault (/07-Agent-Work)"]
    E <-->|Relative Links| F

Track A: Technical Specifications (Docs-as-Code)

  • Format: Standard Markdown files stored directly inside the code repository under the design/ folder.
  • Target Audience: Core developers, external technical reviewers, systems integrators, and automated agents.
  • Key Characteristics:
  • Emphasizes formal accuracy, complete API tables, parameter types, explicit configuration keys, and performance complexity bounds.
  • Follows strict versioning in Git alongside the source code.
  • Captures every edge case and error state.

Track B: ADHD-Friendly / Executive Visual Specifications

  • Format: Highly styled Markdown files stored in the central Obsidian Vault.
  • Target Audience: Shahin Mohammadi, multidisciplinary scientists, busy operators, and onboarding developers.
  • Key Characteristics:
  • Visual Primacy: Integrates extensive Mermaid diagrams, flowcharts, data flows, and state machines to explain logic without heavy blocks of text.
  • Color System: Uses headings and accents styled with colors from the Cytognosis fluorescent dye palette. These colors include Violet (#8B3FC7), Azure (#3B7DD6), Magenta (#E0309E), Indigo (#5145A8), Coral #F26355, and Teal (#14A3A3).
  • Structured Alerts: Highlights important context, tips, and warnings using GitHub-style alerts.
  • Scannability: Features quick-reference lookup tables, bulleted checklists, bold key terms, and copy-pasteable terminal commands.
  • Minimal Prose: Avoids dense paragraphs. We limit prose explanation to 3 sentences max before introducing a visual element.

2. Core Technical Document Types

The Cytognosis documentation system consists of six standardized technical document types. Each type serves a specific purpose in the software development lifecycle.

Architecture Decision Record (ADR)

  • Trigger: You make a significant, hard-to-reverse technical choice.
  • What It Captures: Context, constraints, the bold decision sentence, alternatives considered (at least two), and negative or positive consequences.
  • Rule: Keep ADRs append-only and focused on one decision. Create a new ADR if a decision changes.

Module Specification (Module Spec)

  • Trigger: You build or refactor an importable software module.
  • What It Captures: Public API parameters, return types, raises tables, environment configurations, and dependency lists.
  • Rule: Write specs so a developer can integrate with the module by reading only the spec document.

Request for Comments (RFC)

  • Trigger: You propose a multi-module feature or breaking schema change requiring team consensus.
  • What It Captures: Motivation, current state, detailed proposal design, migration strategy, trade-off matrix, and implementation timeline.
  • Rule: Write and approve the RFC before writing a single line of code.

Research Evaluation (EVAL)

  • Trigger: You compare multiple libraries, tools, or third-party APIs.
  • What It Captures: Scoring matrix with weighted criteria, reproducible setup steps, and real test outcomes.
  • Rule: Base all scoring on hands-on testing. Document failure modes explicitly.

Troubleshooting Guide (TROUBLESHOOT)

  • Trigger: An error occurs more than once or requires non-obvious debugging.
  • What It Captures: Exact error logs in code blocks, step-by-step diagnostic commands, root causes, and prevention strategies.
  • Rule: Group and name issues by the symptom what the user actually sees, not the internal cause.

Changelog (CHANGELOG)

  • Trigger: You cut a release or hit a major project milestone.
  • What It Captures: A clean list of user-facing changes grouped by semantic categories (Added, Changed, Deprecated, Removed, Fixed, Security).
  • Rule: Maintain the Unreleased section as you code. Include migration guides for breaking changes.

3. Document Lifecycle and Complexity Matrix

The following table classifies our documentation assets by their development stage, complexity level, and target audience.

Document Type Development Stage Primary Focus / Capture Complexity Level Primary Use Case Primary Audience
Research Evaluation (EVAL) Research & Ideation Compares tools; records scoring matrices; documents setup failures Moderate Choosing a graph database or benchmarking an LLM API Research Lead, Technical Architect
Request for Comments (RFC) Planning & Design Proposes system-wide features; defines migration paths; lists trade-offs High Pitching a new decentralized biosensing pipeline Entire Engineering Team
Architecture Decision Record (ADR) Architecture Setup Documents hard-to-reverse choices; captures negative trade-offs Moderate to High Deciding to store genomic graphs using LinkML Developers, Compliance Auditors
Module Specification (Module Spec) Implementation Details public classes/APIs; lists parameters; diagrams module relations High Implementing the semantic search client interface Software Engineers, QA Engineers
Changelog (CHANGELOG) Release Grouped changes; version history; migration snippets Low Reviewing what changed between v0.5.0 and v0.6.0 End Users, Project Managers
Troubleshooting Guide (TROUBLESHOOT) Operations & Support Exact error patterns; diagnostic checks; copy-pasteable resolutions Moderate Resolving GCP deployment permission errors SRE, On-call Engineers

4. Visual Callout System (ADHD-Friendly Track)

For our visual track, we use distinct callouts to break up prose and emphasize critical insights. We avoid stacking these boxes consecutively to keep the interface clean.

[!NOTE] Use this callout for background context, structural explanations, and minor implementation details.

[!TIP] Use this callout for performance optimizations, best practices, and CLI productivity shortcuts.

[!IMPORTANT] Use this callout for critical design patterns, mandatory integration rules, and requirements that must be followed.

[!WARNING] Use this callout for breaking changes, potential schema corruption risks, and API deprecations.

[!CAUTION] Use this callout for actions that risk data loss, secure key exposures, or compliance violations.


5. Vault Organization and Connected Graph Standards

We store all ADHD-friendly documents in a structured hierarchy inside our central Obsidian Vault. Every document must act as a node in a connected graph to avoid orphaned assets.

ObsidianVault/
├── 00-Index.md                        ← The central entry point
├── 01-Strategy/                       ← Mission, platforms, blindspots
├── 02-Architecture/                   ← Master plans, system graphs
│   ├── adrs/                          ← ADHD-friendly ADR visual summaries
│   └── rfcs/                          ← Visual proposal flows
├── 03-Modules/                        ← Visual module interaction diagrams
├── 04-Evaluations/                    ← Visual technology matrix cards
├── 05-Troubleshooting/                ← Quick-reference symptom cards
├── 06-Changelogs/                     ← Version release timelines
└── 07-Agent-Work/                     ← Current execution notes and validation logs

Reference Linking Standards

  • Bi-directional Links: Every subdirectory has an index file linking to all child files. Each child file must link back to its parent index.
  • Docs-to-Code Links: Track A files inside Git must contain a footer link pointing to the corresponding Track B file in the Obsidian vault (using obsidian:// URLs).
  • Code-to-Docs Links: Track B files in the vault must link directly to the Git repository folder or the exact source files using file:/// URLs.
  • No Orphans: Any new document must be immediately added to 00-Index.md or a related parent index before completing the task.