GITHUB AGENT SKILLS / 技能收录
给你的 AI 助手
装上专业技能
我们从 GitHub 持续收录、审核并分类优质的 Agent Skills,按职业与使用场景整理,复制一行命令即可安装到 Claude Code、Cursor 等智能体。
按职业浏览 Agent Skills
为不同岗位挑选的高频技能
按使用场景浏览 Agent Skills
从你要完成的任务出发
按创作者浏览 Agent Skills
按 GitHub Stars 排序的优秀创作者
Microsoft
@microsoftopenclaw
@openclawAnthropic
@anthropicsOpenAI
@openaiThe Apache Software Foundation
@apacheGitHub
@githubAutoGPT
@Significant-GravitasGarry Tan
@garrytan全部 Skills 482556
agent-memory
Use when a project's CLAUDE.md has grown past what anyone reads and you want the agent to learn durable facts from its own sessions instead — or when asking why the agent keeps re-learning the same correction, why a remembered rule is wrong, or where a memory line came from. Implements a four-tier store (L0 transcripts / L1 candidates / L2 project context / L3 stable persona) where promotion is earned by recurrence across sessions and days, never by one confident statement, and nothing reaches a committed file without a human adopting it.
agent-launcher-orchestrator
Use when a user wants to build, launch, grade, or schedule a Claude Managed Agent (CMA) in their own Anthropic account — "build me an agent", "launch this as a managed agent", "run this on a schedule", "grade my agent against a rubric", "set up a nightly worker". Reads the per-session goal (./my-agent/goal.json), routes deterministically to one of five phase sub-skills (interview → stage-launch → grade-iterate → run-without-you → wrap-up) via goal_router.py, and compiles the goal+phase into an execution shape (single-pass workflow / bounded grade→iterate loop / recurring cron deployment loop) via loop_compiler.py. Forks context so heavy intake (build sheets, payloads, eval cases) stays out of the parent thread. All launches are emitted as BYOK curl the user runs with their own key; no tool makes API calls. Inspired by anthropics/launch-your-agent (Apache-2.0). Distinct from engineering/agent-harness (generic domain loop) and engineering/write-a-skill (authors Claude Code skills, not CMAs).
agent-eval
Head-to-head comparison of coding agents (Claude Code, Aider, Codex, etc.) on custom tasks with pass rate, cost, time, and consistency metrics. Use when choosing between coding agents, or when a change to an agent setup needs measured pass rate, cost, and time rather than an impression.
adversarial-reviewer
Adversarial code review that breaks the self-review monoculture. Use when you want a genuinely critical review of recent changes, before merging a PR, or when you suspect Claude is being too agreeable about code quality. Forces perspective shifts through hostile reviewer personas that catch blind spots the author's mental model shares with the reviewer.
ad-creative
When the user needs to generate, iterate, or scale ad creative for paid advertising. Use when they say 'write ad copy,' 'generate headlines,' 'create ad variations,' 'bulk creative,' 'iterate on ads,' 'ad copy validation,' 'RSA headlines,' 'Meta ad copy,' 'LinkedIn ad,' or 'creative testing.' This is pure creative production — distinct from paid-ads (campaign strategy). Use ad-creative when you need the copy, not the campaign plan.
ab-test-setup
When the user wants to plan, design, or implement an A/B test or experiment. Also use when the user mentions "A/B test," "split test," "experiment," "test this change," "variant copy," "multivariate test," "hypothesis," "conversion experiment," "statistical significance," or "test this." For tracking implementation, see analytics-tracking.
zoologist
Think and work like an expert Zoologist. Use when a task calls for Zoologist judgment. Reasons from Bauplan, homoplasy, and voucher-backed ICZN taxonomy through COL/WoRMS/GBIF IPT, VertNet/Arctos curation, Folmer COI/BOLD BIN, geomorph GPA, IQ-TREE phylogenetics, and Distance/MARK occupancy–abundance models while treating trap selectivity, checklist drift, barcode-only species, and morphometric digitizing noise as first-class failure modes.
x-ray-synchrotron-scientist
Think and work like an expert X-ray / Synchrotron Scientist. Use when a task calls for X-ray / Synchrotron Scientist judgment. Reasons from photon-matter cross sections, reciprocal-space Q, absorption edges, and absorbed dose through pyFAI, GSAS-II, Athena/Artemis, BornAgain, and foil/silver-behenate calibration while treating fluorescence self-absorption, substrate Bragg misindexing, beam damage and radiation-induced reduction, and Fourier-termination ripples as first-class failure modes.
wind-engineering-specialist
Think and work like an expert Wind Engineering Specialist. Use when a task calls for Wind Engineering Specialist judgment. Reasons from ABL exposure, ASCE 7-22/EN 1991-1-4 wind actions, ASCE 49 BLWT Method 3, rigid vs flexible G/Gf, MWFRS vs C&C, and DAD directionality; treats enclosure GCpi, short-fetch exposure, and aeronautical-tunnel misuse as first-class failure modes.
wildlife-biologist
Think and work like an expert Wildlife Biologist. Use when a task calls for Wildlife Biologist judgment. Reasons from detectability-aware abundance (Distance/mrds, MARK/RMark CJS, secr/oSCR, unmarked occupancy), Camtrap DP and Movebank/amt telemetry, ASM/MBTA/ESA permitting, and MIEM/FAIRe eDNA while treating index-effort bias, closure violation, bait aggregation, apparent-survival emigration, and collar/fix pseudoreplication as first-class failure modes.
wetland-scientist
Think and work like an expert Wetland Scientist. Use when a task calls for Wetland Scientist judgment. Reasons from the three-parameter delineation test (1987 Manual + Regional Supplements, NWPL, FISM), Cowardin/NWI/LLWW mapping, HGM function and Level 1–2–3 RAM, compensatory mitigation/RIBITS, and wetland carbon–GHG flux while treating legacy hydrology, 50/20 cover errors, relict redox nodules, NWI≠jurisdiction, and short well records as first-class failure modes.
welding-joining-engineer
Think and work like an expert Welding & Joining Engineer. Use when a task calls for Welding & Joining Engineer judgment. Reasons from heat input, t8/5, HAZ metallurgy, and restraint/shrinkage through AWS D1.1 prequalified vs qualified WPS, ASME IX/ISO 15614 PQR essential variables, RT/UT acceptance (static vs cyclic), FSW wormhole/kissing-bond windows, and neutron/XRD/hole-drilling residual stress while treating prequalification overreach, planar-UT mis-disposition, and cold-FSW root bonds as first-class failure modes.
weed-scientist
Think and work like an expert Weed Scientist. Use when a task calls for Weed Scientist judgment. Reasons from the weed seed bank, population dynamics, and herbicide mode-of-action biology through log-logistic dose-response (GR50/GR90 in R drc), replicated RCB field trials with susceptible checks, molecular resistance assays (ALS sequencing, EPSPS copy number), and HRAC/WSSA-based MOA rotation while treating drift and carryover, tank-mix antagonism and water-quality failures, and late-escape seed rain as first-class failure modes.
water-resources-scientist
Think and work like an expert Water Resources Scientist. Use when a task calls for Water Resources Scientist judgment. Reasons from hydrologic-cycle mass and energy balance, green-versus-blue water, and nonstationarity through MODFLOW, SWAT/HEC-RAS, WEAP, Budyko closure, and multi-objective KGE/NSE calibration while treating equifinality, unaccounted return flows and stream depletion, single-drought-year safe yield, and efficiency-rebound effects as first-class failure modes.
water-resources-engineer
Think and work like an expert Water Resources Engineer. Use when a task calls for Water Resources Engineer judgment. Watershed hydrology through HEC-HMS into HEC-RAS floodplain and stormwater BMP design — catchment balance, DSS coupling, FEMA products, and quantity/quality detention with defensible calibration.
volcanologist
Think and work like an expert Volcanologist. Use when a task calls for Volcanologist judgment. Reasons from mush reservoirs, volatile exsolution, and conduit fragmentation through WOVOdat/GVP unrest synthesis, MultiGAS–DOAS CO₂/SO₂, melt-inclusion thermobarometry, GACOS/ERA5 InSAR, LP/VLP/VOISS-Net seismology, BET_EF probabilistic forecasting, and LaMEVE recurrence while treating atmospheric InSAR artefacts, MI H₂O diffusion loss, open-vent gas misread, and deterministic eruption countdowns as first-class failure modes.
vlsi-chip-design-engineer
Think and work like an expert VLSI / Chip Design Engineer. Use when a task calls for VLSI / Chip Design Engineer judgment. Reasons from PPA tradeoffs, timing slack, on-chip variation, and foundry rule decks through MCMM STA with OCV/POCV in PrimeTime/Tempus, UPF power intent, SpyGlass/JasperGold CDC, and Calibre DRC/LVS while treating clock-domain crossings, post-CTS hold violations, IR drop, and TT-only signoff as first-class failure modes.
viticulturist-enologist
Think and work like an expert Viticulturist / Enologist. Use when a task calls for Viticulturist / Enologist judgment. Reasons from source–sink canopy balance, terroir as water/nitrogen-mediated ripening, sugar–acid–phenolic trajectories, glucophilic AF and Oenococcus MLF, molecular SO₂ at pH, FOSS/OIV analytics, ISO 4120/QDA sensory, and NDVI selective harvest while treating stuck ferment (YAN/fructose), smoke glycoside release, pH-blind SO₂, and mineral-terroir folklore as first-class failure modes.