RenderCV Testing Context
Gives AI agents RenderCV's official test-authoring standards so every test file, name, and structure follows the project's conventions automatically.
This skill is a purely advisory testing-convention document: it executes nothing, requests no permissions, touches no sensitive data and has no external side effects, so least privilege and data-flow transparency are inherently satisfied. Deducted for: no source attribution or statement of relation to sibling skill files, and no explicit note that confirmation/rollback are not applicable.
In-file examples are self-consistent, naming rules are unambiguous and the mirror-structure mapping is clear. Deducted for: the referenced @tests/ and docs/developer_guide/testing.md were not provided in evidence, so reference reachability and consistency of examples with the current codebase (e.g. the compute_date_string signature) cannot be statically verified.
Trigger condition is clear (use when writing or reviewing tests) and the audience (RenderCV contributors) is explicit. Deducted for: no declared non-fit boundaries, no Chinese-language support notes, and limited direct value to FollowSkills end users (resume generation), with trigger precision evidenced only by the description field itself.
Well-structured with progressive disclosure via external references and sufficient examples. Deducted for: no per-skill versioning, changelog or maintenance responsibility; governance relies on the repository (MIT license, active CI badges), and the publisher is unverified. Name and description match content.
As a coding standard it plausibly improves consistency of contributed tests, a real but indirect benefit. Deducted for: static review cannot verify that tests written under this standard are better; no comparative evidence over generic pytest best practice; output usability not execution-verified.
Rules can be cross-checked against actual test files and the repo's CI/coverage badges. Deducted for: no execution in this review, referenced files not attached, and correspondence between examples and the real test suite is inference rather than verified fact.
- This skill targets RenderCV repository contributors, not end users generating resumes; false triggering yields little value for ordinary users.
- Static review executed nothing; reachability of the referenced tests/ and developer docs, and consistency of examples with the current code, are unverified.
- Publisher is unverified; maintenance responsibility and update path are only inferred from overall repository activity.
- The skill carries no version or changelog of its own and may drift as the repository evolves.
What does this skill do, and when should you use it?
rendercv-testing-context is one of six Agent Skills bundled in the RenderCV repository, a resume builder for academics and engineers. It is a pure knowledge skill — no scripts — that encodes the project's testing conventions: how test files mirror the src/rendercv/ source tree, naming rules for tests and classes, when to use pytest parametrize, where shared fixtures belong in conftest.py, and principles like testing in isolation and preferring real behavior over mocking. It activates when an AI agent writes or reviews tests for RenderCV, keeping generated tests consistent with the codebase rather than inventing a style.
The skill injects these conventions into the agent: 1) test files map one-to-one to source files (e.g., src/rendercv/renderer/templater/date.py → tests/renderer/templater/test_date.py); 2) a single test is named test_ + function/class name, multiple tests are grouped in a Test + PascalCase class; 3) @pytest.mark.parametrize handles input variations instead of duplicated similar tests; 4) shared fixtures go in the conftest.py of the closest relevant folder; 5) principles include focused input→output tests, inlining trivial setup, minimal mocking, and naming tests by expected behavior. It also references the tests/ directory and docs/developer_guide/testing.md for deeper context.
- A developer contributing to RenderCV wants the agent to write tests for a new module in the project's existing style
- Reviewing AI-generated test code against a documented convention baseline for naming and structure
- After refactoring a RenderCV source file, having the agent fill in missing cases in the corresponding test file
- A new team member uses the skill to quickly learn how RenderCV's test directory is organized
- An agent working on functions with many input variants (dates, locales) automatically reaches for parametrize
What are this skill's strengths and limitations?
- Conventions come from the official RenderCV repo, matching its real test structure exactly
- Concrete code examples (parametrize, conftest.py layering, naming rules) that can be imitated directly
- Explicit do/don't lists reduce style drift in AI-generated tests
- Plain Markdown knowledge file with no script dependencies — cheap to port
- Scoped to the RenderCV codebase; its src/rendercv/ mirroring convention has no direct value for other projects
- Reference-only — includes no executable scripts or automated validation
- No dedicated evaluation for this skill in the source; the README's promptfoo evaluation covers the collection's generation abilities, not this testing guide
- References @tests/ and docs/developer_guide/testing.md, which only resolve inside the repository
How do you install this skill?
The skill lives at .claude/skills/rendercv-testing-context/ in the rendercv/rendercv repository. The README gives the install command for the whole skill collection: npx skills add rendercv/rendercv-skill (works with any AI agent supporting the skills standard). How to install only this one sub-skill is not documented in the source.
How do you use this skill?
Once installed, the agent loads this context when you ask it to write or review RenderCV tests, per its description ("Use when writing or reviewing tests"). Example prompt: "Write pytest tests for src/rendercv/schema/models/cv/section.py" — the agent should produce a mirrored test path, follow test_/Test naming, and use parametrize. Exact invocation or configuration steps are not documented in the source.