All instances of 'Boreal Bytes' updated to 'borealBytes' (as @borealBytes in narrative context) across issue, PR, kanban, and SKILL.md. Files: issue-00000050, pr-00000050, SKILL.md
7.3 KiB
Issue-00000050: Add Markdown and Mermaid Writing Skill
| Field | Value |
|---|---|
| Issue | #50 |
| Type | ✨ Feature |
| Severity | 🟡 Medium |
| Priority | P1 |
| Reporter | Clayton Young (borealBytes / Superior Byte Works, LLC) |
| Assignee | Unassigned |
| Date reported | 2026-02-19 |
| Status | Open |
| Users affected | All scientific skill users producing documentation, reports, or diagrams |
| Revenue impact | Indirect — improves skill quality, adoption, and downstream reproducibility |
| Resolved in | PR-00000050 |
| Time to resolution | TBD |
📋 Summary
Create a new skill — markdown-mermaid-writing — that establishes markdown with embedded Mermaid diagrams as the DEFAULT and canonical format for scientific documentation output. The core principle: text-based diagrams are the source of truth. Python-generated or AI-generated images are downstream conversions, never the starting point.
Customer impact
| Dimension | Assessment |
|---|---|
| Who's affected | Researchers, scientists, data engineers using skills for documentation |
| How many | All users producing reports, papers, analysis docs, or any diagrams |
| Business impact | Reduces downstream rendering failures; improves reproducibility and AI-readability of outputs |
| Workaround exists | Yes — users can continue using existing image generation methods |
💡 Motivation and origin
The conversation that sparked this
This feature request emerged from a discussion in the K-Dense Discord on 2026-02-19 about balancing AGENTS.md vs skills and the importance of documentation standards for sharing and combining scientific skills1 .
"One note I might offer you all in the claude-scientific-skills that would make it even better is better directions for writing markdown. Specifically using mermaid diagrams extensively and using tables and those instead of python created diagrams when it makes sense. From my experience, the more you get your reports and files in .md in just regular text, which mermaid is as well as being a simple 'script language'. This just helps with any downstream rendering and especially AI generated images (using mermaid instead of just long form text to describe relationships < tokens). Additionally mermaid can render along with markdown for easy use almost anywhere by humans or AI."
— Clayton Young (@borealBytes), K-Dense Discord, 2026-02-19
The K-Dense team responded positively and confirmed this direction aligns with their upcoming reorganization plans2 .
Why markdown + Mermaid as source
| Aspect | Mermaid in Markdown | Python / AI Images |
|---|---|---|
| Version control | ✅ Full git diff | ❌ Binary blob |
| Token efficiency | ✅ Less than prose descriptions | ❌ Base64 or external refs |
| Human editable | ✅ Any text editor | ❌ Requires regeneration |
| AI parseable | ✅ Structured text, no vision model | ❌ Requires vision model |
| Platform support | ✅ GitHub, GitLab, Notion, VS Code | ⚠️ Requires image hosting |
| Downstream flexibility | ✅ Convert to images anytime | ❌ One-way |
| Accessibility | ✅ accTitle / accDescr on every diagram | ❌ Alt-text often missing |
Three-phase workflow
flowchart LR
accTitle: Documentation Production Workflow
accDescr: Mermaid in markdown is the required source phase. Python and AI-generated images are optional downstream phases for polished output only.
source["📄 Phase 1 — Mermaid in Markdown<br/>(ALWAYS — source of truth)"]
python["🐍 Phase 2 — Python Generated<br/>(optional — when charts needed)"]
ai["🎨 Phase 3 — AI Generated<br/>(optional — polished visuals)"]
final["📊 Final Output"]
source --> python
source --> ai
python --> final
ai --> final
source --> final
classDef source fill:#dbeafe,stroke:#2563eb,stroke-width:2px,color:#1e3a5f
classDef optional fill:#fef9c3,stroke:#ca8a04,stroke-width:2px,color:#713f12
classDef output fill:#dcfce7,stroke:#16a34a,stroke-width:2px,color:#14532d
class source source
class python,ai optional
class final output
🎯 Acceptance criteria
Must have
- New skill directory:
scientific-skills/markdown-mermaid-writing/ SKILL.mdwith YAML frontmatter, philosophy section, workflow guide- Markdown style guide (
references/markdown_style_guide.md) — full guide, ~730 lines - Mermaid style guide (
references/mermaid_style_guide.md) — full guide, ~454 lines - All 24 diagram type guides (
references/diagrams/*.md) - All 9 document templates (
templates/*.md) - Example research report in
assets/examples/ - Issue and PR documents following the AGENTS.md "Everything is Code" convention
Must not have
- Modification of any existing skills
- Breaking changes to repo structure
- Removal of Python/AI image generation options — those remain valid as downstream phases
Integration points
scientific-schematics— use after mermaid source to generate AI-polished imagesscientific-writing— use this skill for all document structure and diagrams within manuscriptsliterature-review— use this skill for formatting findings and relationship diagrams
🔍 Technical details
Content sources
All style guides and templates ported from the SuperiorByteWorks-LLC/agent-project repository under Apache-2.0 license with attribution retained3 .
| Source path | Destination |
|---|---|
agentic/markdown_style_guide.md |
references/markdown_style_guide.md |
agentic/mermaid_style_guide.md |
references/mermaid_style_guide.md |
agentic/mermaid_diagrams/*.md (24 files) |
references/diagrams/*.md |
agentic/markdown_templates/*.md (9 files) |
templates/*.md |
Skill directory structure
scientific-skills/markdown-mermaid-writing/
├── SKILL.md
├── references/
│ ├── markdown_style_guide.md
│ ├── mermaid_style_guide.md
│ └── diagrams/ ← 24 diagram type guides
│ ├── flowchart.md
│ ├── sequence.md
│ └── [22 more]
├── templates/ ← 9 document templates
│ ├── pull_request.md
│ ├── issue.md
│ └── [7 more]
└── assets/
└── examples/
└── example-research-report.md
📝 References
Last updated: 2026-02-19
-
K-Dense Discord thread, 2026-02-19. Discussion between Clayton Young (@borealBytes), Orion Li (K-Dense), and Timothy about AGENTS.md standards and scientific documentation. ↩︎
-
Orion Li (K-Dense), responding: "Thanks for letting us know! We'll make sure to include this into our sauce when we implement our upcoming reorganization of the skills." — K-Dense Discord, 2026-02-19. ↩︎
-
Boreal Bytes / Superior Byte Works, LLC.
opencoderepository. https://github.com/SuperiorByteWorks-LLC/agent-project. Apache-2.0 License. ↩︎