architecture-vocabulary
Skill14 starsChanged 8 months agoMITView on GitHub
Use when evaluating architecture in tech-review — provides formal vocabulary (Depth, Seam, Leverage, Locality, Adapter, Pass-through) with diagnostic heuristics that produce precise, falsifiable findings instead of generic descriptions
In this repository50 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- adr-conventionSkill
- backlog-externalSkill
- backlog-localSkill
- backlogSkill
- checkpointsSkill
- design-principlesSkill
- domain-glossarySkill
- git-practicesSkill
- grounded-claimsSkill
- knowledge-checkSkill
- policy-riskSkill
- positive-framingSkill
- receiving-code-reviewSkill
- skill-authoringSkill
- skill-awarenessSkill
- subagent-driven-developmentSkill
- test-driven-developmentSkill
- token-efficientSkill
- triageSkill
- verification-before-completionSkill
- workflow-bugSkill
- workflow-checkSkill
- workflow-commitSkill
- workflow-contractsSkill
- workflow-debugSkill
- workflow-decisionsSkill
- workflow-docsSkill
- workflow-doctorSkill
- workflow-epicSkill
- workflow-featureSkill
- workflow-flow-fix-pipelineSkill
- workflow-flowSkill
- workflow-grillSkill
- workflow-handoffSkill
- workflow-ideaSkill
- workflow-implementSkill
- workflow-planSkill
- workflow-proposalSkill
- workflow-prSkill
- workflow-refineSkill
- workflow-researchSkill
- workflow-reviewSkill
- workflow-startSkill
- workflow-statusSkill
- workflow-tech-reviewSkill
- workflow-update-workflowSkill
- workflow-validateSkill
- workflow-worktreeSkill
- workflow-yoloSkill
The file
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 8
Use these terms and heuristics when evaluating architecture.
- Distinctive: line 23
When forming architecture findings, follow this protocol:
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Architecture Vocabulary
Use these terms and heuristics when evaluating architecture. Every architecture finding should reference at least one vocabulary term. Replace generic language with precise claims.
Terms
| Term | Definition | Diagnostic Heuristic |
| Depth | Deep module: simple interface, rich functionality. Shallow module: interface nearly as complex as its implementation. | Count public methods vs lines of meaningful logic. Ratio near 1:1 = shallow. |
| Seam | A point where behavior can be swapped without changing surrounding code. | One adapter = hypothetical seam. Two adapters = real seam. Zero = no seam yet. |
| Adapter | A thin translation layer at a seam boundary. Contains no business logic. | If your adapter has business logic, it's a service pretending to be an adapter. |
| Leverage | How much behavior a module provides per unit of interface complexity. | Deletion test: imagine deleting the module. If complexity reappears across callers, it was earning its keep (high leverage). If callers barely notice, it was unnecessary indirection (low leverage). |
| Locality | Related changes stay together. Modifying one behavior touches one file, not five. | A single conceptual change requiring edits in 3+ files = poor locality. The concept is scattered across wrong boundaries. |
| Pass-through | A method or layer that receives arguments, adds nothing, and forwards them. | If you can describe it as "calls X with the same arguments," it's a pass-through. Adds stack depth and maintenance cost without value. |
Applying the Vocabulary
When forming architecture findings, follow this protocol:
- Identify which term applies to the observation
- Measure using the term's diagnostic heuristic
- State the finding as a falsifiable claim with evidence
Findings: Before and After
Before: "This module has poor separation of concerns." After: "This module is shallow — 14 public methods, each a thin wrapper over the database client (ratio ~1:1). Consider consolidating into fewer, deeper methods."
Before: "The code is tightly coupled." After: "No seam exists between the notification service and the email provider (zero adapters). Adding a second provider would require modifying the service internals."
Before: "There's unnecessary abstraction here."
After: "The EventDispatcher has low leverage — deletion test shows callers would each add one direct call, replacing the dispatcher's 200-line interface with ~3 lines each. The indirection isn't earning its complexity."
Before: "Changes to billing touch too many files."
After: "Billing logic has poor locality — updating the tax calculation requires edits in models/order.ts, services/billing.ts, utils/tax.ts, and api/checkout.ts (4 files). Consider co-locating tax logic."
Integration
This skill is loaded by:
commands/tech-review.md— used during the Architecture & Structure evaluation dimension
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

