tech-docs
Generate professional Architecture Documents, Software Design Documents (SDD/TDD), and Architecture Decision Records (ADRs) using SOTA templates based on…
Skill by AbdelazizMoustafa10m
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/abdelazizmoustafa10m-tech-docs
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: documents, design, testing, writing
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/tech-docs
What's inside
- Skill name
tech-docs- Description
- Generate professional Architecture Documents, Software Design Documents (SDD/TDD), and Architecture Decision Records (ADRs) using SOTA templates based on C4+arc42, Google-style design docs, and Structured MADR formats. Use this skill whenever the user wants to create, write, or generate any technical documentation including: architecture documents, system design docs, software design documents, technical design documents, design docs, ADRs, decision records, RFCs, or technical specifications. Also trigger when the user says things like "document this decision", "write up the architecture", "create a design doc for this feature", "I need an ADR", or asks to document technical decisions, system design, or architecture. Even if they just say "document this" in a technical context, this skill likely applies.
- License
- MIT
- Instructions
- 105 lines in SKILL.md · view on GitHub
Files (4)
- SKILL.md
- adr-template.md · references/
- architecture-template.md · references/
- design-doc-template.md · references/
SKILL.md (first lines)
# Tech Docs: Architecture, Design, and Decision Documentation
You are generating professional technical documentation. This skill supports three document types,
each serving a distinct purpose in the documentation hierarchy:
| Document | Purpose | Audience | Lifecycle |
|---|---|---|---|
| **Architecture Document** | System-wide structure, boundaries, and major choices | Architects, leads, new joiners | Stable; updated when structure changes |
| **Design Document (SDD/TDD)** | Implementation plan for a specific feature/change | Engineers building the feature + reviewers | Change-driven; written before implementation |
| **ADR** | Capture the "why" behind a single architectural decision | Current team + future maintainers | Immutable once accepted; superseded, never deleted |
## Step 1: Determine Document Type
If the user hasn't specified, determine from context:
- Talking about a **specific decision** (database choice, framework, pattern) → **ADR**
- Planning a **feature, service, or major change** → **Design Document**
- Describing the **overall system** or onboarding documentation → **Architecture Document**
Ask if ambiguous: "Would you like an ADR (single decision), a Design Doc (feature plan), or an Architecture Document (system overview)?"
## Step 2: Gather Context
Before generating, understand the project:
1. **Read the codebase** — scan the project structure, README, existing docs, and any `docs/adr/` or `docs/design/` directories to understand the system
2. **Check for existing ADRs** — if generating an ADR, find the next sequential number
3. **Ask the user** about anything you can't determine from code:
- For ADRs: What decision needs to be made? What options are being considered?
- For Design Docs: What feature/change? What problem does it solve?
- For Architecture Docs: What system? What level of detail (MVP, production, comprehensive)?
## Step 3: Generate the Document
Read the appropriate reference template and generate the document:
- **ADR** → Read `references/adr-template.md` for full template guidance
- **Design Document** → Read `references/design-doc-template.md` for full template guidance
- **Architecture Document** → Read `references/architecture-template.md` for full template guidance
### Key principles for all document types
**Trade-offs over prescriptions.** The most valuable part of any technical document is the explicit
statement of what was gained and what was sacrificed. Every design choice involves trade-offs —
document them honestly. "We chose X, accepting Y as a consequence" is more useful than "We chose X
because it's better."
**Why before What.** Always explain the problem and motivation before presenting the solution.
A reader who understands why a decision matters can evaluate whether the chosen approach still
makes sense as circumstances change.
**Concrete and measurable.** Prefer "p99 latency < 200ms" over "fast." Prefer "supports 50K
concurrent users" over "scalable." Quantified targets make documents actionable and reviewable.
**Diagrams as code.** Use Mermaid syntax for all diagrams so they render in GitHub/GitLab and
are version-controllable. Every architecture doc should have at least a C4 Context diagram and
a Container diagram.
**Docs-as-code.** Documents live in the repo alongside the code they describe. Store ADRs in
`docs/adr/` (or `decisions/`), design docs in `docs/design/`, and the architecture doc at
`docs/architecture.md`. This makes them discoverable, reviewable in PRs, and versioned.
## Step 4: File Placement
Suggest appropriate file placement based on project conventions:
```
docs/
├── architecture.md # System architecture (one per system)
├── design/ # Feature/change design docs
│ ├── YYYY-MM-DD-feature-name.md
│ └── ...
└── adr/ # Decision records (immutable log)
├── 001-decision-title.md
├── 002-decision-title.md
└── ...
```
If the project already has a different convention (e.g., `decisions/`, `docs/decisions/`), follow
the existing pattern.
## ADR Format Selection
ADRs come in multiple formats. Default to **Structured MADR** but adapt based on context:
| Format | When to use | Complexity |
|---|---|---|
| **Structured MADR** (default) | Teams needing traceability, automation, or compliance | High |
| **MADR** | Good balance of structure and simplicity for most teams | Medium |
| **Classic Nygard** | Simple human-readable records, minimal overhead | Low |
| **Y-statement** | Ultra-fast capture of tactical, low-risk decisions | Minimal |
If the user says "quick ADR" or "simple ADR", use Classic Nygard or Y-statement format.
If the user mentions compliance, audit, or machine-readable, use Structured MADR.
## Architecture Document Detail Levels
Scale the architecture document to the project's maturity:
- **MVP / Prototype**: Executive Summary + System Context (C4 L1) + Key ADRs only
- **Production Service**: Add Building Blocks (C4 L2-L3) + Deployment + Cross-Cutting Concerns
- **Complex System**: Full template with all sections
Ask the user or infer from the codebase which level is appropriate.From AbdelazizMoustafa10m/skills-for-agents, MIT.
More from AbdelazizMoustafa10m/skills-for-agents
- 0.570score
codex-counsel
AbdelazizMoustafa10m
Consult OpenAI's Codex CLI (a different model family; default model gpt-6-astra) for an independent, read-only second opinion.
SkillAFreshness: AClaude Code +9 - 0.570score
codex-sol
AbdelazizMoustafa10m
Delegate bounded research, analysis, codebase scouting, or audits to GPT-6.1 Sol through Codex CLI in Fast mode.
SkillAFreshness: AClaude Code +9 - 0.570score
unslop
AbdelazizMoustafa10m
Rewrite or review human-facing prose to remove generic AI mannerisms while preserving meaning, facts, uncertainty, citations, intended voice, and project…
SkillAFreshness: AClaude Code +9 - 0.570score
writing-for-agents
AbdelazizMoustafa10m
Write and revise documents that an agent reads as instructions, including skills, AGENTS.md, CLAUDE.md, system prompts, tool and MCP descriptions, subagent…
SkillAFreshness: AClaude Code +9 - 0.570score
codex-implementer
AbdelazizMoustafa10m
Delegate hands-on implementation to OpenAI's Codex CLI (a GPT-family coding worker; default model gpt-6-astra) while Claude stays the manager: Claude writes a…
SkillAFreshness: AClaude Code +9 - 0.567score
emil-design-eng
AbdelazizMoustafa10m
Apply Emil Kowalski-inspired UI restraint and component polish when that style is requested, or consult a focused supporting recipe.
SkillAFreshness: AClaude Code +9 - 0.565score
orchestrate
AbdelazizMoustafa10m
Coordinate specialist workers and reviewers while the main session manages scope, handoffs, and acceptance.
SkillAFreshness: AClaude Code +9 - 0.565score
commit-message
AbdelazizMoustafa10m
Draft conventional commit messages from staged changes.
SkillAFreshness: AClaude Code +9 - 0.541score
teach
AbdelazizMoustafa10m
Teach the user a new skill or concept, within this workspace.
SkillAFreshness: AClaude Code +9 - 0.539score
handoff
AbdelazizMoustafa10m
Compact the current conversation into a concise handoff for another agent.
SkillAFreshness: AClaude Code +9 - 0.307score
frontend-craft
AbdelazizMoustafa10m
No description from the publisher yet.
SkillAFreshness: AClaude Code +9 - 0.307score
design-motion-principles
AbdelazizMoustafa10m
No description from the publisher yet.
SkillAFreshness: AClaude Code +9
Signals
- Stars
- 0
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Oct 3, 2026 (25 days ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from AbdelazizMoustafa10m
- 0.570score
codex-counsel
AbdelazizMoustafa10m
Consult OpenAI's Codex CLI (a different model family; default model gpt-6-astra) for an independent, read-only second opinion.
SkillAFreshness: AClaude Code +9 - 0.570score
codex-sol
AbdelazizMoustafa10m
Delegate bounded research, analysis, codebase scouting, or audits to GPT-6.1 Sol through Codex CLI in Fast mode.
SkillAFreshness: AClaude Code +9 - 0.570score
unslop
AbdelazizMoustafa10m
Rewrite or review human-facing prose to remove generic AI mannerisms while preserving meaning, facts, uncertainty, citations, intended voice, and project…
SkillAFreshness: AClaude Code +9 - 0.570score
writing-for-agents
AbdelazizMoustafa10m
Write and revise documents that an agent reads as instructions, including skills, AGENTS.md, CLAUDE.md, system prompts, tool and MCP descriptions, subagent…
SkillAFreshness: AClaude Code +9 - 0.570score
codex-implementer
AbdelazizMoustafa10m
Delegate hands-on implementation to OpenAI's Codex CLI (a GPT-family coding worker; default model gpt-6-astra) while Claude stays the manager: Claude writes a…
SkillAFreshness: AClaude Code +9 - 0.567score
emil-design-eng
AbdelazizMoustafa10m
Apply Emil Kowalski-inspired UI restraint and component polish when that style is requested, or consult a focused supporting recipe.
SkillAFreshness: AClaude Code +9
Cite this page
tech-docs skill. Agent Mart. https://agentmart.dev/skills/abdelazizmoustafa10m-tech-docs (last updated 2026-10-24)