Research

Research, evidence gathering, literature, reports, investigation, and synthesis

skills in category: 30,315
pages available: 1,264

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 research skills

Showing 11,041–11,064 of 30,315 skills

Research ManagerAResearch project skeleton and file lifecycle. Load when initializing a `.kilo/` project, migrating or archiving documents, or deciding where to put a new research note. Defines the four-directory layout (proposal/project/archive/knowledge), two-digit numeric prefixes, the global.md index layer, archive SUMMARY.md rules, and final REPORT.md requirements. It does not run experiments (experiment_manager) or judge whether an idea is worth pursuing (research_progress).researchgogitVotes: 0GitHub stars: 5
Knowledge KeeperAKnowledge-base retrieval, citation-graph expansion, and capture for papers and external sources. Load when the user asks to search literature, look up a paper, summarize a paper into the knowledge base, or reuse previously found references. Enforces local-first retrieval, forward citation chasing from key papers, a query log, and mandatory capture. Judging whether a literature gap is real or a direction is worth doing belongs to research_progress.researchapiVotes: 0GitHub stars: 5
Academic Paper WritingATop-tier CS/AI/Robotics manuscript writing taste. Load when drafting, revising, or polishing academic papers in computer vision, machine learning, robotics, image processing, and related fields. Not for research reports or slides — those belong to write_md and slide_deck.researchVotes: 0GitHub stars: 5
Security AuditBAudit a diff (a release tag range or a PR branch) for anything malicious or a serious security regression before shipping/merging. Use when asked to "check this release", "make sure nothing malicious slipped in", "security review this PR", or vet upstream/contributor changes. Covers supply chain, egress/exfiltration, auth/injection/traversal, CI/infra, and commit provenance, with a fast direct pass plus fanned-out deep review. Trigger on any "is this change safe?" request over a diff.researchpythonrustVotes: 0GitHub stars: 13,244
Pr ReadyARun the pre-submit gauntlet (security-audit, code-review, simplify, review-polish, humanizer) over a change before opening or merging a PR.researchrustgoVotes: 0GitHub stars: 13,244
Knowledge BaseARead Greptile's synthesized knowledge base for this repository through the Greptile MCP (list_knowledge_bases, list_knowledge_base_documents, get_knowledge_base_document, search_knowledge_base). Use whenever you want context about the codebase, such as how a subsystem or flow works, where something lives, its architecture and conventions, or before planning a change in modules you have not read yet. Covers the lookup order, search and truncation limits, verifying claims against the checkout, ...researchrustapiVotes: 0GitHub stars: 13,244
Perf LoopAOrchestrates the three-layer performance loop (bench → load → diagnose → optimize → verify) on the local Aspire-hosted backend. Use when the user asks to "run the perf loop", "load test X", "profile the backend", "where do the allocations come from", "why is this operation slow", or to verify a performance fix end-to-end. Routes to the right layer and tool; does not duplicate them. For pure BenchmarkDotNet before/after comparison use `perf-benchmark` directly.researchrustgoVotes: 0GitHub stars: 2
Qa MonitoringA[QA Method] Online bug monitoring from Application Insights: query both layers, dedup by fingerprint, triage new/spiking signatures, reproduce HIGH-confidence bugs live, report. Detect-and-report only — never files a bug tracker ticket or auto-fixes.researchgosqlVotes: 0GitHub stars: 2
Qa InvestigateA[QA Method] Use when a suspected or triaged bug needs a live reproduction, layer isolation and a root cause BEFORE it is filed — reproduce, isolate, gather evidence (an evidence package with evidence-index.md + root-cause.md), common VC patterns. The step between triage (/qa-triage-results Phase 4) and /vc-fix:qa-bug, which reuses its package.researchgoazureVotes: 0GitHub stars: 2
Qa Triage ResultsA[QA Methodology] Triage a completed regression run's non-passing cases (FAIL / BLOCKED / SKIPPED) — or, with `ticket <KEY>`, the non-passing rows of a /qa-test or /qa-test-fast testing-checklist.md (checklist, exploratory and visual findings): collect each + its evidence (trace, screenshots, console, network) + the failing CSV or checklist row, dedup + flag flaky, classify each into real-product-bug vs test-defect (bad steps / bad assertion / stale test data / stale test) vs flaky/env/known, ...researchrailstestingVotes: 0GitHub stars: 2
Qa Review BlA[QA Method] Pipeline entry point for the BL sync — ALIAS of /qa-review-oracles bl. Called automatically by /qa-test-lifecycle Phase 4c on the BL-* a run contradicted or surfaced (mark SUSPECT, resolve from the Jira decision, sync from the ticket's AC or docs; edits bl/<slug>.yaml only from a human source; reconcile test-case citations). For a manual oracle review, use /qa-review-oracles.researchVotes: 0GitHub stars: 2
Qa Perf MeasureA[QA Method] Measure backend work per request on a DEPLOYED environment and prove whether a change moved it: dependency calls by type (search/SQL/cache/HTTP) via an App Insights operation_Id join, N+1 detection by input-size scaling, and paired positive/negative controls so a null result is trustworthy. Counts transfer between environments; latency does not. Measure-and-report only.researchrustgoVotes: 0GitHub stars: 2
Qa MonitoringA[QA Method] Online bug monitoring from Application Insights: query both layers, dedup by fingerprint, triage new/spiking signatures, reproduce HIGH-confidence bugs live, report. Detect-and-report only — never files JIRA or auto-fixes.researchgosqlVotes: 0GitHub stars: 2
Qa InvestigateA[QA Method] Bug investigation: reproduce, isolate root cause, gather evidence, common VC patterns.researchgogitVotes: 0GitHub stars: 2
Guru Review Change RequestAReview whether a current change request is one independently deliverable unit through linked prerequisite evidence, an AI readiness gate, and five typed exits.researchVotes: 0GitHub stars: 8
Guru Qualify Normal ScenarioAQualify one profile-specific candidate set against current normal-scenario authority before a Guru Team caller may promote it into scope, tests, findings, implementation, or publication blocking.researchVotes: 0GitHub stars: 8
Guru Maintain Architecture BaselineAMaintain a repository Architecture Baseline through semantic bootstrap, task impact synchronization, promotion, and repair profiles.researchVotes: 0GitHub stars: 8
Guru Execute Task Free ChangeAExecute one bounded task-free change through AI-owned checkout suitability, editing, targeted validation, and risk-evolution review.researchgoVotes: 0GitHub stars: 8
Evm Wallet Docker E2eARun the evm-wallet Docker e2e tests (build, start stack, wait for healthy, test, diagnose failures).researchbashdockerVotes: 0GitHub stars: 9
Duck WhyAName the cause of a failure before anyone writes a fix, because the symptom is not the defect. Use when a test fails, a bug is reported, a build or CI goes red, a traceback or error is pasted, something passes in one environment and fails in another, behavior differs from what was expected, a fix keeps not sticking, or the user asks why something is broken, including when the code is pasted inline and the cause looks obvious. Also when the user asks who added a value, flag, guard or behavior ...researchpythongoVotes: 0GitHub stars: 5
Duck ReviewAReview work from multiple angles or deliver an independent judgment. Use for designs, plans, documents, code, local changes, completed work or PRs; findings-only reviews; an independent second opinion; a release gate; or to answer a review you received, saying which comments still apply to what you pushed since and drafting the replies. Findings-only review stays in-session; independent and release reviews use cross-model scrutiny and return APPROVE, REJECT, or NOTE. It does not fix, repeat, ...researchpythonrustVotes: 0GitHub stars: 5
Duck RaceAPut two different model families on the same problem and let executed evidence pick the result. Use when the user says "race it", "duck race", or "ping-pong", wants two models tackling one problem, a task has several plausible implementations worth comparing, generated tests keep passing without catching real defects, single-attempt builds of similar work kept failing review, or a review-fix loop keeps faulting the fixes instead of the original change.researchpythonrustVotes: 0GitHub stars: 5
Duck ProofAMake the goal, the path, and the finished work earn your trust through counterexamples and executed checks. Use when the user asks whether a solution is necessary, correct, or elegant, asks to prove a plan or finished shape, or wants work verified before review.researchrustgoVotes: 0GitHub stars: 5
Duck DietAPut agent context, memory, and token costs on a diet without starving the essential guidance. Use when the user asks for minimum tokens, a session-cost analysis, an agent setup health check, or trimming of CLAUDE.md, AGENTS.md, or memory, when picking which model tier or agent type a stage runs on, or before a long campaign or multi-agent run.researchsecurityVotes: 0GitHub stars: 5