architecture-docs
Where architecture documents live, what each contains, and the conformance report format.
Skill by Dev-Toolbelt
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/dev-toolbelt-architecture-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
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 and code. 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/architecture-docs
What's inside
- Skill name
architecture-docs- Description
- Where architecture documents live, what each contains, and the conformance report format.
- License
- MIT
- Instructions
- 79 lines in SKILL.md · view on GitHub
Files (1)
- SKILL.md
SKILL.md (first lines)
# Architecture Documents Canonical location, content contract, and conformance reporting format for the documents an architect produces. Load when writing, updating, or validating any of them. --- ## Document Location Rule — CRITICAL **Every architecture document is written to the path the doc-layout map resolves for its key.** `scripts/lib/doc-layouts.json` is canonical — this table mirrors it for reading convenience; when they disagree, the JSON wins. Resolve a path with `bash .dev-team-agents/scripts/doc-path.sh <key>` (empty output means the project's layout has no such doc). Paths are relative to the docs write root (`docs/` by default). Layout rules: `skills/shared/doc-flow/SKILL.md` § Doc Layout. | Key | `development` layout (default) | `gated` layout | |-----|-------------------------------|----------------| | `trd` | — | `docs/trd.md` | | `architecture` | `docs/development/architecture.md` | `docs/architecture.md` | | `tech-stack` | `docs/development/tech-stack.md` | `docs/tech-stack.md` | | `code-standards` | `docs/development/code-standards.md` | `docs/code-standards.md` | | `data-model` | `docs/development/database.md` | `docs/data-model.md` | | `api-contracts` | `docs/development/api-contracts.md` | `docs/api-contracts.md` | | `threat-model` | — | `docs/security/threat-model.md` | | `adr` | `docs/development/adrs/NNNN-*.md` | `docs/adr/ADR-NNNN-*.md` | | `evals` | — | `docs/ai/evals.md` | | `testing` | `docs/tests/testing-strategy.md` | `docs/testing.md` | | `conformance-report` | `docs/development/conformance-report.md` | `docs/conformance-report.md` | The ADR directory is always whatever `new-adr.sh --dir` prints (`skills/shared/adr/SKILL.md`). **NEVER write architecture documents to:** - `.opencode/` — opencode configuration only - `.claude/` — Claude Code configuration only - `.dev-team-agents/` — the framework installation directory - Any hidden directory (starting with `.`) --- ## Content Contract per Document | Key | Must contain | |-----|--------------| | `trd` | Index of the technical documents only — one link per doc with a one-line purpose. No rule-level detail. | | `architecture` | System design, layers, component boundaries, integration patterns, and the layer depth chosen **per domain area**. Links to `data-model` and `api-contracts`; never repeats them. Every significant decision also gets an ADR. | | `tech-stack` | Chosen technologies with rationale, including what was considered and **rejected**. | | `code-standards` | Patterns, naming conventions, linting rules, and design patterns — specific to this project's stack, not generic advice. | | `data-model` | **Data Model Specification**: database choice, logical overview, one Mermaid `erDiagram` per domain area, data dictionary (table, column, type, constraint, meaning), PII tags per column, migration approach, indexing guidelines. Template: `.dev-team-agents/templates/data-model-template.md`. Finalize only after `database-specialist` review. | | `api-contracts` | API style (REST/GraphQL/RPC), versioning policy, auth model, response and error envelope, every endpoint. | | `threat-model` | Threats per trust boundary and the personal-data map, from `security-specialist` findings. Template: `.dev-team-agents/templates/threat-model-template.md`. | | `evals` | Only when the product uses LLMs: eval cases, metrics and pass thresholds per AI task. | | `testing` | Testing strategy — levels, tooling, fixtures, what runs where. | **Owner rule** — one fact, one owning doc (`skills/shared/doc-flow/SKILL.md` § Owner Rule): schema facts live only in `data-model`, endpoints only in `api-contracts`. `architecture` and `trd` link to them by section or ID, never repeat them. **Spec self-review before handing off** — silently check every document for: placeholders and TODOs, contradictions between documents, ambiguous decisions, YAGNI violations, and scope creep. Fix inline before asking the user to review. --- ## Conformance Report Produced or updated after each Quality Gate run. Flag every finding with one tag: | Tag | Meaning | |---|---| | `[ARCH-DEVIATION]` | The code does not follow the decided architecture | | `[TECH-DEBT]` | Acceptable shortcut, but it must be tracked | | `[CONFORMANT]` | This is correct | Write to the `conformance-report` doc (path from the table above). **This file accumulates across sprints — append, never overwrite.** ```markdown ## Conformance Report — [Sprint / Date] ### Summary [Overall conformance posture — 1-2 sentences] ### Deviations | File / Area | Deviation | Severity | Action | |---|---|---|---| | service/OrderService.js | Business logic in controller | HIGH | Refactor before next sprint | ### Tech Debt Tracked | Item | Introduced | Owner | Target Sprint | |---|---|---|---| ### Conformant Areas [List areas that were reviewed and passed] ```
From Dev-Toolbelt/dev-team-agents, MIT.
More from Dev-Toolbelt/dev-team-agents
- 0.623score
design-patterns
Dev-Toolbelt
Design patterns — KISS, YAGNI, DRY, SOLID, GoF, DDD reference.
SkillAFreshness: A★ 7Claude Code +9 - 0.620score
graphql
Dev-Toolbelt
GraphQL — schema, resolvers, N+1/DataLoader, pagination.
SkillAFreshness: A★ 7Claude Code +9 - 0.618score
api-design
Dev-Toolbelt
REST/GraphQL design — naming, status codes, pagination, idempotency.
SkillAFreshness: A★ 7Claude Code +9 - 0.618score
api-versioning
Dev-Toolbelt
API versioning — compatibility rules and deprecation lifecycle.
SkillAFreshness: A★ 7Claude Code +9 - 0.618score
async-jobs
Dev-Toolbelt
Async jobs — idempotency, retry strategy, DLQ, observability.
SkillAFreshness: A★ 7Claude Code +9 - 0.618score
i18n
Dev-Toolbelt
i18n/l10n — locale detection, translations, pluralization, RTL.
SkillAFreshness: A★ 7Claude Code +9 - 0.618score
component-patterns
Dev-Toolbelt
Component patterns — container/presentational for React/Vue/Angular.
SkillAFreshness: A★ 7Claude Code +9 - 0.618score
css-quality
Dev-Toolbelt
CSS quality — specificity, magic numbers, responsive.
SkillAFreshness: A★ 7Claude Code +9 - 0.618score
data-fetching-integrity
Dev-Toolbelt
Detect and prevent duplicate/redundant API calls in SPA/SSG frontends — tool-agnostic.
SkillAFreshness: A★ 7Claude Code +9 - 0.618score
event-driven
Dev-Toolbelt
Event-driven — sourcing, CQRS, saga, idempotency, schema registry.
SkillAFreshness: A★ 7Claude Code +9 - 0.618score
form-handling
Dev-Toolbelt
Form design — state, validation, error feedback.
SkillAFreshness: A★ 7Claude Code +9 - 0.618score
frontend-patterns
Dev-Toolbelt
Frontend patterns — debounce, double-submit prevention, error boundaries.
SkillAFreshness: A★ 7Claude Code +9
Signals
- Stars
- 7
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Oct 18, 2026 (10 days ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/Dev-Toolbelt/dev-team-agents
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from Dev-Toolbelt
- 0.623score
design-patterns
Dev-Toolbelt
Design patterns — KISS, YAGNI, DRY, SOLID, GoF, DDD reference.
SkillAFreshness: A★ 7Claude Code +9 - 0.620score
graphql
Dev-Toolbelt
GraphQL — schema, resolvers, N+1/DataLoader, pagination.
SkillAFreshness: A★ 7Claude Code +9 - 0.618score
component-patterns
Dev-Toolbelt
Component patterns — container/presentational for React/Vue/Angular.
SkillAFreshness: A★ 7Claude Code +9 - 0.618score
async-jobs
Dev-Toolbelt
Async jobs — idempotency, retry strategy, DLQ, observability.
SkillAFreshness: A★ 7Claude Code +9 - 0.618score
api-design
Dev-Toolbelt
REST/GraphQL design — naming, status codes, pagination, idempotency.
SkillAFreshness: A★ 7Claude Code +9 - 0.618score
api-versioning
Dev-Toolbelt
API versioning — compatibility rules and deprecation lifecycle.
SkillAFreshness: A★ 7Claude Code +9
Cite this page
architecture-docs skill. Agent Mart. https://agentmart.dev/skills/dev-toolbelt-architecture-docs (last updated 2026-10-24)