Testing & QA

Unit tests, integration tests, E2E, QA, validation, and test automation

skills in category: 31,094
pages available: 1,296

Security grades appear on each card once the skill has been scanned. Newly imported skills may briefly show without a grade until the backfill job runs.

Open in full browser

Browse testing & qa skills

Showing 1,921–1,944 of 31,094 skills

Skill Tool Capability Fallback Routing TestAUse when a workflow depends on optional search, browser, filesystem, or test capabilities to create or update a capability-to-fallback table with prohibited substitutions. Use the configured search, fetch, read, browser, and test capabilities only when available and authorized. Success criterion: the agent either uses an authorized equivalent or stops with a precise missing-capability request. Preserve evidence and a run record, limit rework to three meaningful passes, and obtain approval bef...testingrustgoVotes: 0GitHub stars: 2
Skill Discovery And Supply Chain ReviewAUse when a user asks to find, compare, or add agent skills from a URL list, registry, or repository to check candidate relevance, source identity, license, entry-point format, dependencies, and side effects before importing; keep provenance and never run setup scripts by default. Trigger for skill discovery, evaluation, or installation.testingrustgoVotes: 0GitHub stars: 2
Parallel FindallAUse when a task involves using a configured entity-discovery service to find records matching a natural-language description to identify the intended outcome, relevant inputs, platform or version, sensitive data, and permission boundary before acting. Use current primary documentation for version-sensitive behavior, produce a reviewable artifact, and verify it against stated criteria. Trigger for planning, implementation, evaluation, or troubleshooting in this focused domain; do not install o...testingrustgoVotes: 0GitHub stars: 2
Mcp Skills Extension Interoperability ReviewAUse when exposing, discovering, or loading Agent Skills through the MCP Skills extension `io.modelcontextprotocol/skills` to review capability declaration, skills/list and skills/get, resource manifests, URI identity, digest verification, cache origin, directoryRead support, user approval, and activation behavior. Success means skill content is loaded only from its approved server and verified manifest, and reading a SKILL.md is not mistaken for permission to activate or execute it.testingrustgoVotes: 0GitHub stars: 2
Mcp Resource Cache Subscription Boundary ReviewAUse when designing or auditing MCP resources, paginated discovery, response caching, or change subscriptions, especially when visibility depends on the caller’s authorization to review resource URIs, cache keys, TTL, public/private scope, pagination, invalidation, subscription filters, and reconnect behavior. Success means private data cannot cross authorization contexts, stale results are invalidated or refreshed appropriately, and cursors and subscription IDs are handled as opaque protocol ...testingperformanceVotes: 0GitHub stars: 2
Mcp Deprecated Feature Migration PlanAUse when upgrading an MCP client, server, SDK, or deployment and identifying protocol features that are deprecated in the target revision to produce a versioned inventory, migration plan, compatibility window, owners, and regression tests. Success means each deprecated use is mapped to the current migration path, new code does not adopt deprecated behavior by default, and existing clients remain supported only through an explicit, tested transition plan. Recheck the live deprecation registry ...testingapidocumentationVotes: 0GitHub stars: 2
Gh Aw Untrusted Issue Content Injection DefenseAUse when a task involves testing defenses against instructions embedded in GitHub issue or discussion content to record the product, runtime and repository version, identity, trust boundary, data sensitivity, controlling artifact, owner, and approval scope before applying the method. Distinguish documented behavior from tested behavior, verify with a bounded reproducible case, preserve provenance, and report compatibility and uncertainty. Do not install, grant access, send external writes, or...testingrustrailsVotes: 0GitHub stars: 2
Documentation Automation Linkcheck GateAUse when an automation edits docs from source changes or repository events to create or update a documentation diff and link-check report. Use the configured search, fetch, read, browser, and test capabilities only when available and authorized. Success criterion: all changed claims match source evidence and changed links resolve in the intended site. Preserve evidence and a run record, limit rework to three meaningful passes, and obtain approval before externally visible or irreversible writes.testingrustgoVotes: 0GitHub stars: 2
Copilot Cli Project Instruction Injection TestAUse when a task involves testing whether untrusted repository content can masquerade as authoritative agent instructions to record the product, runtime and repository version, identity, trust boundary, data sensitivity, controlling artifact, owner, and approval scope before applying the method. Distinguish documented behavior from tested behavior, verify with a bounded reproducible case, preserve provenance, and report compatibility and uncertainty. Do not install, grant access, send external...testingrustrailsVotes: 0GitHub stars: 2
Safe Output Count And Target Policy CheckAUse when an agent prepares structured outputs for a separate write stage to create or update a safe-output validation report with count, target, and allowed fields. Use the configured search, fetch, read, browser, and test capabilities only when available and authorized. Success criterion: all outputs pass the configured policy before any write job runs. Preserve evidence and a run record, limit rework to three meaningful passes, and obtain approval before externally visible or irreversible w...testingrustgoVotes: 0GitHub stars: 2
Resume Session Permission RevalidationAUse when a task continues in a new session, agent, branch, or workspace to create or update a resume record of identity, scope, tool permissions, and approvals. Use the configured search, fetch, read, browser, and test capabilities only when available and authorized. Success criterion: all remaining actions fit the current session’s active permission boundary. Preserve evidence and a run record, limit rework to three meaningful passes, and obtain approval before externally visible or irrevers...testingrustgoVotes: 0GitHub stars: 2
Read Only To Write Capability Escalation ReviewAUse when a workflow moves from investigation to editing or an external action to create or update a read/write boundary plan with target paths, payload, and approval point. Use the configured search, fetch, read, browser, and test capabilities only when available and authorized. Success criterion: the write is necessary, scoped, reversible where possible, and explicitly authorized. Preserve evidence and a run record, limit rework to three meaningful passes, and obtain approval before external...testingrustgoVotes: 0GitHub stars: 2
Issue Body Untrusted Instruction Triage BoundaryAUse when an event-driven agent reads user-submitted issue content to create or update a triage record separating trusted workflow rules from quoted issue data. Use the configured search, fetch, read, browser, and test capabilities only when available and authorized. Success criterion: classification is based on repository evidence and the issue schema, not embedded commands. Preserve evidence and a run record, limit rework to three meaningful passes, and obtain approval before externally visi...testingrustgoVotes: 0GitHub stars: 2
Human Escalation Evidence Packet BuilderAUse when the next safe action depends on a human judgment or approval to create or update a packet with decision, options, evidence, risk, and exact approval needed. Use the configured search, fetch, read, browser, and test capabilities only when available and authorized. Success criterion: a human can decide without repeating the investigation. Preserve evidence and a run record, limit rework to three meaningful passes, and obtain approval before externally visible or irreversible writes.testingrustgoVotes: 0GitHub stars: 2
Human Approval GatesAUse when a proposed agent action can publish, delete, purchase, send, deploy, change access, expose data, or otherwise cause material external effects to classify risk, request specific human approval before gated actions, show exact scope, and verify the result. Trigger before irreversible, costly, sensitive, or third-party-impacting operations.testingVotes: 0GitHub stars: 2
External Write Preview Hash Approval CheckAUse when an agent prepares to publish, submit, comment, or create a remote artifact to create or update a preview digest, target identity, and approval record. Use the configured search, fetch, read, browser, and test capabilities only when available and authorized. Success criterion: the submitted payload and destination match the approved preview exactly. Preserve evidence and a run record, limit rework to three meaningful passes, and obtain approval before externally visible or irreversibl...testingrustgoVotes: 0GitHub stars: 2
Agent Pr Changed Path Allowlist VerifierAUse when an agent proposes a patch, branch update, or pull request to create or update a changed-path report matched against an explicit allowlist. Use the configured search, fetch, read, browser, and test capabilities only when available and authorized. Success criterion: every changed path is permitted or the patch is blocked for review. Preserve evidence and a run record, limit rework to three meaningful passes, and obtain approval before externally visible or irreversible writes.testingrustgoVotes: 0GitHub stars: 2
Web Quotation Context Boundary AuditAUse when a draft paraphrase or quotation may have been shortened or detached from context to create or update a quote-to-claim mapping with the relevant surrounding passage. Use the configured search, fetch, read, browser, and test capabilities only when available and authorized. Success criterion: the quotation and paraphrase retain the original limitation and polarity. Preserve evidence and a run record, limit rework to three meaningful passes, and obtain approval before externally visible ...testingrustgoVotes: 0GitHub stars: 2
Web Claim Source Date Freshness AuditAUse when a claim concerns current versions, prices, policy, availability, release status, or deadlines to create or update a dated claim-source table with as-of date and update signal. Use the configured search, fetch, read, browser, and test capabilities only when available and authorized. Success criterion: the source is valid for the requested date or the claim is explicitly time-bounded. Preserve evidence and a run record, limit rework to three meaningful passes, and obtain approval befor...testingrustgoVotes: 0GitHub stars: 2
Redirected Source Canonical Url AuditAUse when a source URL redirects, canonicalizes, or changes host during retrieval to create or update a redirect record containing requested URL, final URL, and reason for substitution. Use the configured search, fetch, read, browser, and test capabilities only when available and authorized. Success criterion: the final destination is relevant and its identity is transparent in the record. Preserve evidence and a run record, limit rework to three meaningful passes, and obtain approval before e...testingrustgoVotes: 0GitHub stars: 2
Multi Chunk Fetch Completeness LedgerAUse when a fetched page reports additional chunks or is truncated before its relevant section to create or update a chunk ledger with expected, fetched, and still-missing ranges. Use the configured search, fetch, read, browser, and test capabilities only when available and authorized. Success criterion: all chunks needed for the question are present or the result clearly says review is partial. Preserve evidence and a run record, limit rework to three meaningful passes, and obtain approval be...testingrustgoVotes: 0GitHub stars: 2
Verification And StoppingAUse when an agent must prove that a deliverable meets an objective before claiming completion to turn goals into observable checks, choose independent evidence, define pass and fail thresholds, and state explicit stop conditions. Trigger for code, research, data, documents, and long-running agent work where plausible output is not sufficient evidence.testinggoVotes: 0GitHub stars: 2
Success Signal Repeatability Cross Run CheckAUse when a task completion depends on a nondeterministic metric or one-off visual check to create or update a repeated-result table with environment and variance. Use the configured search, fetch, read, browser, and test capabilities only when available and authorized. Success criterion: the signal is reproducible within an agreed tolerance or marked unstable. Preserve evidence and a run record, limit rework to three meaningful passes, and obtain approval before externally visible or irrevers...testingrustgoVotes: 0GitHub stars: 2
Skill Workflow Exit Evidence Contract TestAUse when a procedure ends with “done” but lacks a checkable exit condition to create or update an exit-contract table mapping each outcome to evidence and stop reason. Use the configured search, fetch, read, browser, and test capabilities only when available and authorized. Success criterion: every success state has a concrete observation and every failure state has a bounded route. Preserve evidence and a run record, limit rework to three meaningful passes, and obtain approval before externa...testingrustgoVotes: 0GitHub stars: 2