Solid Skills — Senior-Engineer Coding Skills for AI Agents
Turns junior-level code into senior-engineer quality software through SOLID principles, TDD, and clean architecture.
The skill is instruction-only content with no observed scripts or network access, limiting attack surface, so most trust criteria get 1 rather than 0; however the README claims it 'automatically detects and fixes code smells' with no user-confirmation gate and no rollback guidance (rollback 0). Attribution is partly handled: Academind credit and MIT declaration earn 2 for source_attribution.
README is internally consistent (skill name, structure, install command all align), scoring 2; but it depends on an external Agent Skills runtime, installation is a single npx path, and failure behavior is undocumented, each scored 1.
Audience (TypeScript/NestJS) and use scenarios are explicit, scoring 2; trigger conditions are overly broad ('writing any code'), capability boundaries and unsuitable environments are unstated, each deducted to 1.
Structure, naming, install notes and MIT license claim are clear (2s); but no LICENSE file evidence, no changelog, no known-limitations section, and no maintenance/update path — all deducted.
Bundling nine reference documents gives real marginal value (2); but SKILL.md and the reference files are not in evidence, so output usability is speculative, and context-size cost of the references is undiscussed (1s).
Core claims (methods under 10 lines, automatic smell fixing) carry no attached evidence and the cited files are absent from review; facts and inferences are not separated; only the Academind credit is externally checkable — 1 across the board.
- Only the README was reviewable; SKILL.md and the nine reference files are absent from evidence, so actual skill content, quality, and size cannot be verified.
- The skill claims to modify code automatically with no user-confirmation or rollback mechanism documented; manually review every code change before production use.
- No version number, changelog, or maintenance commitment; update and regression risk falls on the user.
- MIT is only asserted in README text; confirm a LICENSE file exists in the repository.
What does this agent do, and when should you use it?
Solid Skills is a set of professional software engineering skills for AI coding agents, published at github.com/ramziddin/solid-skills in the Agent Skills format defined by anthropics/skills. It ships one skill, called solid, which enforces a TDD workflow and SOLID principles whenever the agent writes, refactors, reviews, or tests code. The skill consists of a SKILL.md instruction file plus nine reference documents covering SOLID principles, TDD, testing strategies, clean code, code smells, GoF design patterns, clean architecture, object design, and complexity management. It is primarily designed for TypeScript and NestJS projects but claims applicability to any object-oriented codebase. Once installed, skills are automatically available and used when the agent detects relevant tasks. The repository is MIT licensed.
After installing with npx skills add ramziddin/solid-skills, the solid skill injects a set of coding directives: enforce the TDD Red-Green-Refactor cycle (failing test first), apply the five SOLID principles to every class and function, detect and fix code smells automatically, use value objects for domain primitives (IDs, emails, money), follow the Law of Demeter and Tell Don't Ask, keep methods under 10 lines and classes under 50 lines. The nine references files (solid-principles.md, tdd.md, testing.md, clean-code.md, code-smells.md, design-patterns.md, architecture.md, object-design.md, complexity.md) provide the agent with detailed, TypeScript-example-backed documentation.
- A TypeScript/NestJS developer who wants the agent to write the failing test first when implementing a feature such as user registration.
- An engineer maintaining a legacy codebase who wants a service refactored into single-responsibility, dependency-inverted structures.
- A team reviewer who wants the agent to detect code smells and propose fixes during code review.
- An architect designing a payment-system module who wants design guidance based on vertical slicing and the dependency rule.
- A developer adding tests to an existing module who wants strategy and pattern guidance.
What are this agent's strengths and limitations?
- Ships nine structured reference documents (GoF patterns, clean architecture, complexity management), giving far more depth than a single-file prompt skill.
- Enforces concrete, checkable constraints (methods <10 lines, classes <50 lines, value objects, Law of Demeter) instead of vague "write good code" advice.
- Mandates a TDD-first workflow, changing the agent's default implement-first habit.
- Designed primarily for TypeScript and NestJS; effectiveness on other languages or paradigms is undocumented and requires self-evaluation.
- No tests, releases, or benchmarks are published, so claimed quality improvements lack auditable evidence.
- Requires the host agent to support the Agent Skills (anthropics/skills) format; environments without that support cannot use it directly.
How do you install or deploy this agent?
Run inside an AI coding agent environment that supports the Agent Skills format:
npx skills add ramziddin/solid-skillsRequires Node.js (for the npx command). No other credentials or runtime requirements are documented.
How do you use this agent?
No configuration is needed; skills are automatically available once installed and used when relevant tasks are detected. Example prompts: "Implement a user registration feature", "Refactor this service to follow SOLID principles", "Review this code for quality issues", "Add tests for this module", "Design the architecture for a payment system".
How does this agent compare with similar options?
The repository follows the Agent Skills format from Anthropic's official skills repo (anthropics/skills) and installs as a third-party skill in that ecosystem; no competing products are named in the README.