0196
frontend-design
Design constraints that push generated interfaces beyond generic AI UI.
- Main limitation
- Aesthetic judgments remain subjective and still need stakeholder review.
- First risk
- Style over function · Medium
- Source
- Pinned · 0a64e398
AGENT SKILLS DIRECTORY
Code Review & Testing Agent Skills
Find Skills that help inspect changes, design tests, challenge assumptions, and verify software work before it ships.
- 33
- indexed Skills
- 20
- Top 100 entries
- 3
- native Agent signals
JOBS TO BE DONE
Start with the outcome
- 01Review code and architecture decisions
- 02Generate and maintain focused tests
- 03Add evidence to development handoffs
SELECTION CHECKS
Reduce the decision before install
- 01Confirm which files and commands the Skill may access.
- 02Separate review guidance from tools that can modify code.
- 03Run the first test in a disposable branch or workspace.
THE COMPLETE CATEGORY
Code review & testing
Ordered by SkillSignal's published metadata signals. A score is not a security certification—open the full profile before installing.
0294
web-design-guidelines
Review interfaces against practical web design and accessibility rules.
- Main limitation
- It depends on network access and a live guideline document whose contents can change over time.
- First risk
- Rule drift or context-free findings · Medium
- Source
- Pinned · b8caa260
0394
mcp-builder
Plan, implement, and evaluate high-quality MCP servers in Python or TypeScript.
- Main limitation
- The guide cannot replace current MCP SDK or target-API documentation.
- First risk
- Overpowered external-service tools · High
- Source
- Pinned · 0a64e398
0494
skill-creator
Create, improve, and benchmark Agent Skills with trigger tests and comparative evaluation.
- Main limitation
- Meaningful benchmarks require representative prompts, repeat runs, and careful human judgment.
- First risk
- Untrusted evaluation prompts and artifacts · High
- Source
- Pinned · 0a64e398
0593
grill-me
Relentless decision interviews for sharper plans and designs.
- Main limitation
- A relentless interview can be tiring and disproportionate for small, reversible decisions.
- First risk
- Decision fatigue · Low
- Source
- Pinned · 885e2ca4
0693
codeql
Interprocedural security analysis with explicit database-quality and query-suite gates.
- Main limitation
- Requires CodeQL, jq, uv, a buildable target, and potentially long database construction.
- First risk
- Build and scan side effects · High
- Source
- Pinned · 9b281335
0793
semgrep
Approval-gated Semgrep security scans with telemetry disabled and auditable SARIF output.
- Main limitation
- Requires Semgrep and network access for registry or third-party rules; Pro-only cross-file analysis may be unavailable.
- First risk
- Third-party rules and source exposure · High
- Source
- Pinned · 9b281335
0893
differential-review
Risk-first security review for diffs, commits, and pull requests with blast-radius analysis.
- Main limitation
- Not intended for greenfield code or broad initial vulnerability discovery.
- First risk
- False assurance from incomplete context · High
- Source
- Pinned · 9b281335
0993
autoresearch
Autonomous, metric-driven code experiments that keep improvements and discard regressions.
- Main limitation
- Only fits tasks with a stable, representative numeric metric.
- First risk
- Autonomous repository mutation · High
- Source
- Pinned · 318066d2
1093
anti-ui-slop
Product-specific UI design contracts grounded in real interface evidence and a hard finish gate.
- Main limitation
- Requires access to the repository and ideally external UIZZE references or user-provided screenshots.
- First risk
- Copying external interface patterns too literally · Medium
- Source
- Pinned · 318066d2
1193
bigquery-pipeline-audit
A targeted Python and BigQuery audit for runaway cost, unsafe reruns, scan size, writes, and observability.
- Main limitation
- Specialized to Python plus BigQuery and does not validate the live project configuration or billing account.
- First risk
- Audit estimates mistaken for billing guarantees · High
- Source
- Pinned · 318066d2
1292
variant-analysis
Systematic hunts for other instances of a known bug or vulnerability root cause.
- Main limitation
- Requires one validated bug in hand; it is not an initial discovery workflow.
- First risk
- Overbroad automated matches · Medium
- Source
- Pinned · 9b281335
1392
web-design-reviewer
Browser-based visual review for responsive, accessible, consistent web interfaces with source-level fixes.
- Main limitation
- Requires a running target, browser automation, and source access for fixes.
- First risk
- Unintended changes during remote review · High
- Source
- Pinned · 318066d2
1492
create-implementation-plan
Machine-readable implementation plans with atomic phases, dependencies, tests, risks, and unique identifiers.
- Main limitation
- Demanding zero ambiguity can create oversized plans or false precision for exploratory work.
- First risk
- False certainty from generated detail · Medium
- Source
- Pinned · 318066d2
1592
create-technical-spike
Time-boxed technical research documents that turn one critical unknown into an evidence-backed decision.
- Main limitation
- Provides a document structure, not proof that experiments are representative or complete.
- First risk
- Prototype work affects real systems · High
- Source
- Pinned · 318066d2
1691
tdd
A practical test-driven development workflow for coding agents.
- Main limitation
- The workflow needs a functioning test runner and clear public behavior.
- First risk
- False confidence · Medium
- Source
- Pinned · 885e2ca4
1791
context-map
A pre-change map of files, dependencies, tests, reference patterns, and implementation risks.
- Main limitation
- The workflow is intentionally brief and may miss dynamic dependencies or runtime ownership.
- First risk
- Incomplete map treated as full scope · Medium
- Source
- Pinned · 318066d2
1890
find-skills
Find Skills is Vercel's discovery workflow for locating and installing reusable Agent Skills.
- Main limitation
- Discovery quality depends on skills.sh coverage and current network access.
- First risk
- Unreviewed installation · Medium
- Source
- Pinned · 435076e7
1990
vercel-react-best-practices
Vercel React Best Practices is a maintained performance guide for React and Next.js.
- Main limitation
- Rules need interpretation against the application's React, Next.js, and deployment versions.
- First risk
- Context-free optimization · Low
- Source
- Pinned · b8caa260
2086
improve-codebase-architecture
Improve Codebase Architecture explores code for architectural friction and proposes deeper modules with smaller interfaces.
- Main limitation
- Architecture quality depends on change history and domain context that may be incomplete.
- First risk
- Premature refactoring · Medium
- Source
- Pinned · 885e2ca4
2186
setup-matt-pocock-skills
Setup Matt Pocock Skills is a repository configuration scaffold for a family of engineering workflows.
- Main limitation
- The output is tailored to Matt Pocock's Skill family rather than a universal standard.
- First risk
- Repository configuration changes · Medium
- Source
- Pinned · 885e2ca4
2286
handoff
Handoff turns the current conversation into a compact continuation document for a fresh agent.
- Main limitation
- Compression can omit nuance that mattered in the original conversation.
- First risk
- Sensitive context leakage · High
- Source
- Pinned · 885e2ca4
2386
triage
Triage moves project issues through a small, configured state machine of triage roles.
- Main limitation
- The workflow requires correctly configured tracker operations and label vocabulary.
- First risk
- External tracker mutation · High
- Source
- Pinned · 885e2ca4
2486
prototype
Prototype creates throwaway code to answer one design or logic question.
- Main limitation
- Prototype code intentionally omits production hardening, persistence, and broad edge-case coverage.
- First risk
- Prototype promoted to production · Medium
- Source
- Pinned · 885e2ca4
2585
teach
Teach turns the current directory into a persistent learning workspace for one topic across many sessions.
- Main limitation
- It is for a sustained learning project, not a one-off explanation or an instant answer.
- First risk
- Untrusted teaching sources or unwanted workspace writes · Medium
- Source
- Pinned · 885e2ca4
2685
grilling
Grilling stress-tests a plan, decision, or idea as a dependency-aware design tree.
- Main limitation
- It deliberately delays implementation until the decision tree is complete and explicitly confirmed.
- First risk
- Decision fatigue or recommendation anchoring · Low
- Source
- Pinned · 885e2ca4
2785
caveman
Caveman is a persistent response-compression mode with lite, full, ultra, and classical-Chinese variants.
- Main limitation
- Compression can omit nuance, ambiguity, or rationale that mattered in the original conversation.
- First risk
- Critical nuance hidden by compression · Medium
- Source
- Pinned · 702da5ce
2885
domain-modeling
Domain Modeling turns a codebase's ubiquitous language into an actively maintained engineering artifact.
- Main limitation
- It documents language and decisions; it does not replace implementation design or domain-expert validation.
- First risk
- Institutionalizing the wrong model · Medium
- Source
- Pinned · 885e2ca4
2985
repo-intake-and-plan
Repo Intake and Plan is a README-first helper for research-code repositories.
- Main limitation
- It does not create environments, download assets, execute commands, perform paper lookup, or write the final report.
- First risk
- Planning from incomplete repository evidence · Medium
- Source
- Pinned · 3ab50525
3085
minimal-run-and-audit
Minimal Run and Audit executes a selected smoke, inference, evaluation, or other short non-training verification for research code.
- Main limitation
- It does not choose the reproduction target, train a model, conduct broad paper analysis, or hide scientific-meaning changes.
- First risk
- Scientific changes hidden as setup fixes · High
- Source
- Pinned · 3ab50525
3185
env-and-assets-bootstrap
Environment and Assets Bootstrap prepares a conservative, Conda-first environment and asset plan after a research reproduction target is already chosen.
- Main limitation
- It assumes the target has already been selected and is not a repository-intake or paper-interpretation workflow.
- First risk
- Untrusted assets or irreproducible dependencies · High
- Source
- Pinned · 3ab50525
3285
codebase-design
Codebase Design supplies a shared vocabulary for reviewing modules, interfaces, seams, adapters, leverage, locality, and testability.
- Main limitation
- The vocabulary helps critique design but does not produce a correct architecture automatically.
- First risk
- Premature abstraction · Medium
- Source
- Pinned · 885e2ca4
3385
diagnosing-bugs
Diagnosing Bugs is an evidence-first loop for difficult defects and performance regressions.
- Main limitation
- It requires a repeatable pass/fail signal; intermittent production-only failures may need observability or human evidence first.
- First risk
- Secrets captured in diagnostic evidence · High
- Source
- Pinned · 885e2ca4