Agent skills you can check before you install
Public SKILL.md skills from GitHub repositories with 50+ stars and from the vendors themselves, scanned and scored. Install any of them with one command in Claude Code, Cursor, Codex, GitHub Copilot or Gemini CLI.
49,409 skills from 2,375 repositories · updated 16 minutes ago
48,362 skills
- minutes-graphsilverstein/minutesPolicy-safe relationship rankings, commitments, aliases, person profiles, and topic research. Always use Minutes' bounded native CLI surfaces; never build or read a durable graph cache.ScienceNo warnings★ 1.6K↓ —Also in 1 other repo
- five-mindsmohitagw15856/pm-claude-skillsAnswer a question five completely different ways — as five independent minds with clashing worldviews — then converge on what survives. Use when asked for multiple perspectives, look at this from every angle, what would different people think, or give me a range of views not one answer. Produces five genuinely distinct takes (each committed to one worldview, not hedged), the tensions between them made explicit, and a final synthesis that keeps what's strongest — deliberately widening the range before narrowing it.OtherNo warnings★ 1.5K↓ —
- demo-scriptmohitagw15856/pm-claude-skillsScript a product demo that lands — the audience's-workflow storyline (their day, not your feature list), the golden path rehearsed with fallbacks, the wow moment placed early, and the demo-death contingencies (the backup video, the reset state, the narration bridge). Use when asked script our product demo, demo this to a customer/exec, our demos meander through features, or the demo broke live last time. Produces the demo storyline, the click-path script with fallbacks, the wow placement, and the contingency kit.MediaNo warnings★ 1.5K↓ —
- feishunano-muse/nanoMusenanoMuse: an open-source personal agent for every device you own. One agent with a name and a look of its own that does things, keeps working while the app is closed, remembers you, and asks before anything you cannot undo. On Android, iPhone and iPad, Windows / macOS / Linux and in the browser; your phone's screen and your computer as its hands.BrowserNo warnings★ 3.1K↓ —
- deck-autopsymohitagw15856/pm-claude-skillsAutopsy a slide deck from photos or screenshots of its slides — the narrative arc, the numbers, and what each slide is hiding. Use when given slide images (a competitor's pitch, a conference talk, your own deck before a big meeting) and asked what the deck argues, whether it holds up, or how to counter or improve it. Produces a slide-by-slide read, the reconstructed argument chain, weak links, and the questions the deck is engineered to avoid. Requires image input.DocumentsNo warnings★ 1.5K↓ —
- agentlas-cloudagentlas-ai/Agentlas-OSUse when the user types /agentlas-cloud, /agentlas cloud, or /hep-cloud to staff only from the signed-in owner's private Cloud packages.OtherNo warnings★ 1.6K↓ —
- hn-digestmohitagw15856/pm-claude-skillsPull the current Hacker News front page, top comments, or a topic search with zero API keys — the official Firebase API and Algolia search via curl, digested instead of dumped. Use when asked what's on Hacker News, summarize HN today, what's the discussion on this story, or has HN covered some topic. Produces a ranked digest with scores and comment counts, the discussion's actual argument threads when asked, and the rerunnable commands.EngineeringNo warnings★ 1.5K↓ —
- oma-docsfirst-fluke/oh-my-agentCheck documentation references, sync docs to code changes, and detect translation drift. Use for documentation maintenance in a repository.EngineeringNo warnings★ 1.4K↓ —
- cold-emailmohitagw15856/pm-claude-skillsWrite a cold sales/B2B outreach email that earns a reply. Use when asked to write a cold email, a sales outreach email, a prospecting email, or a cold email sequence to a business prospect. Produces a short, personalised email — subject, a relevant opener, one clear value-led ask, and a low-friction CTA — plus 2 follow-ups, written to be replied to, not deleted.ProductivityNo warnings★ 1.5K↓ —
- experimentation-analyticsrampstackco/claude-skillsHow to read experiment results without fooling yourself. Confidence intervals, p-values, multiple testing, sequential testing, CUPED, heterogeneous treatment effects, ratio metrics, network effects, dashboard reconciliation, and the interpretation failures that produce confidently wrong shipping decisions. Use this skill whenever the user is reading a finished experiment result panel and about to make a ship, kill, or iterate decision, or when an experiment number does not match the dashboard number. Triggers on read experiment results, result panel, ship or kill decision, p-value, confidence interval, statistical significance, multiple testing, peeking, sequential testing, CUPED, variance reduction, heterogeneous treatment effects, ratio metric, network effects, inconclusive test, experiment versus dashboard mismatch. Use `experiment-design` instead when the test has not run yet and the question is hypothesis, sample size, duration, or what to test.DataNo warnings★ 1K↓ —
- oma-skill-creatorfirst-fluke/oh-my-agentCreate or update OMA skills in the SSL-lite human-readable format. Use when adding a new `.agents/skills/{skill-name}/SKILL.md`, converting an existing skill to the standardized Scheduling / Structural Flow / Logical Operations / References structure, or validating whether a skill has enough routing, execution, resource, and safety detail.OtherNo warnings★ 1.4K↓ —
- electronfcakyon/claude-codex-settingsAutomate Electron desktop apps (VS Code, Slack, Discord, Figma, Notion, Spotify, etc.) using agent-browser via Chrome DevTools Protocol. Use when the user needs to interact with an Electron app, automate a desktop app, connect to a running app, control a native app, or test an Electron application. Triggers include "automate Slack app", "control VS Code", "interact with Discord app", "test this Electron app", "connect to desktop app", or any task requiring automation of a native Electron application.TestingNo warnings★ 1.2K↓ —
- write-reportgooseworks-ai/goose-skillsTurns findings you already have into a report a marketer can read in 30 seconds. It has a short summary with key findings and how sure each one is, the first move, decisions needed, at most six next moves, what was not checked, and dated sources. It checks length and wording before handing over, then saves the report to the brand. Use alone when you already have the findings and their sources; otherwise run the playbook for the job (e.g. research-market, research-competitors, plan-campaign).WritingNo warnings★ 1.3K↓ —
- mono-coloryanliudesign/mono-color-skillGenerate original one-ink or controlled two-ink editorial images from any theme, sentence, article idea, object, or reference photo. Always use this skill when the user asks for 单色海报、双色印刷、单色调视觉、蓝色/绿色孔版印刷、risograph、网点照片、复古或当代编辑排版、zine poster, monochrome editorial poster, duotone print, or asks to use the mono-color style. It uses an adaptive white, gray, or pale-beige substrate, no more than two printing inks, active negative space, terse human language, and strong serif/grotesk/mono typography without making retro styling the default or copying a source composition, wording, logo, or artwork. Produce both the final generation prompt and the generated raster image unless the user explicitly asks for prompt only.DesignNo warnings★ 3.6K↓ —
- dpa-reviewmohitagw15856/pm-claude-skillsRead a Data Processing Agreement before you sign it — sub-processors, transfer mechanism, breach-notice window, deletion, audit rights — in plain language with 🔴🟡🟢 risk. Use when asked to review a DPA, check a data processing agreement, is this DPA safe to sign, or what am I agreeing to on data. Produces the plain-English summary, the risk-ranked findings, the missing-clause checklist, and the questions to send back before signature.EngineeringNo warnings★ 1.5K↓ —
- watchgooseworks-ai/goose-skillsObserve actual video and sound over the whole file or named ranges, with timestamped evidence and honest coverage. Use complementary playback, listening and targeted frames; return observations without edits or verdicts.MediaNo warnings★ 1.3K↓ —
- ctf-forensicsljagiello/ctf-skillsProvides digital forensics and signal analysis techniques for CTF challenges. Use when analyzing disk images, memory dumps, event logs, network captures, cryptocurrency transactions, steganography, PDF analysis, Windows registry, Volatility, PCAP, Docker images, coredumps, side-channel power traces, DTMF audio spectrograms, packet timing analysis, CD audio disc images, or recovering deleted files and credentials.SecurityNo warnings★ 3.6K↓ —
- oma-academic-writingNewfirst-fluke/oh-my-agentDraft and revise academic prose against a rubric, evidence, and citation requirements. Use for essays, reports, literature reviews, or academic style audits.WritingNo warnings★ 1.4K↓ —
- marketing-data-accessgooseworks-ai/goose-skillsHow to get live public data (ads libraries, social posts and comments, search results, creator profiles), business identity/work-contact reads, domain SEO metrics and paid media generation through GooseWorks, with no vendor key. Load before any playbook step that reads public data or calls an outside model: which proxy call fits the job, what it costs, how to get the human's yes, and what to do when the data is not reachable. Not for brand facts (read the brand brain) or the customer's own ad accounts (the GooseWorks ads tools, or an ads connector they have).MarketingNo warnings★ 1.3K↓ —
- doca-ngaugeNewNVIDIA/skillsUse this skill when the operator is driving the DOCA ngauge tool to analyze, visualize, and debug network performance on a single node by probing NIC hardware counters and storing collected data in HDF5 format. Trigger even when the user does not say "DOCA ngauge" — implicit phrasings include "probe NIC hardware counters", "collect HDF5 performance data", "plot ngauge output", "sample counters on BlueField/ConnectX", "measure network performance on a single node", "debug network performance with hardware counters", "run ngauge YAML config", "parse ngauge HDF5 output". Refuse and route elsewhere: DOCA install/bring-up to doca-setup, version questions to doca-version, hardware/firmware state changes to doca-hardware-safety, raw packet I/O to doca-eth, and host-DPU messaging to doca-comch.EngineeringNo warnings★ 3.8K↓ —
- render-song-mvgooseworks-ai/goose-skillsAssemble a song-driven music-video ad from a config — a generated sung track carries the whole narration across N tableaux (one keyframe -> one i2v clip per lyric beat) with NO separate voiceover, captions synced to the song's OWN word timings (script-window, never Whisper) and the hook word landing on the chorus drop, closed on a PIL brand end card. This is the FREE deterministic assembly stage (clip cut-to-timeline + captions + end card + FFmpeg composite); the song, keyframes, and clips come from create-music-elevenlabs / create-image-fal / create-video-fal. Use for the song-driven-music-video format.MediaNo warnings★ 1.3K↓ —
- executive-cvmohitagw15856/pm-claude-skillsWrite an executive or senior leadership CV and the matching board biography: scope, business results and governance up front, two to three pages, written for search firms and boards. Use when asked to write my executive CV, director or C-suite resume, CV for a headhunter, board bio, or non-executive director application. Produces the executive CV with a scope line per role, a results-led profile, a 150-word board biography in the third person, and a short brief for search consultants.WritingNo warnings★ 1.5K↓ —
- where-to-start-situation-savegooseworks-ai/goose-skillsStep of the where-to-start playbook: saves the founder's answers and the agreed first move with the brand, so later visits and other playbooks don't ask again. Run where-to-start instead; not for direct use.OtherNo warnings★ 1.3K↓ —
- icdt-submissionbrycewang-stanford/Awesome-Journal-SkillsUse when auditing an ICDT (International Conference on Database Theory) regular-paper submission for Microsoft CMT readiness, covering the two-step abstract-then-paper deadline in the correct submission cycle, the lipics-v2021 15-page limit excluding references, the clearly-marked appendix read at the PC's discretion, anonymous submission since 2024, the complete-proofs / full-version expectation, and which submission problems are unfixable after the AoE cutoff.DataNo warnings★ 1.4K↓ —
finetuning
evo-hq/evo
DesignUpdated Oct 5, 2026
From the author
This skill should be used when picking or diagnosing a training move (SFT, LoRA, DPO/KTO/ORPO, RFT, GRPO/PPO/RLOO, RLHF), or when the user mentions fine-tuning, post-training, training recipe, reward design, or weight updates. Decision tree by reward shape, smoke-run gate, three failure diagnostics, five false-progress patterns. Provider recipes and I/O contract in references/.
Install
npx skills add evo-hq/evo --skill finetuning -a claude-code
Installs to .claude/skills/finetuning/ · Each command fetches the skill straight from GitHub with the open skills CLI.
Download .zip
Byte-identical to the scanned commit, with the repository’s licence and a SOURCE.txt.
What it can touch
Instructions onlyNo network use seenLicence: APACHE-2.0
Safety checks on SKILL.md
No warnings.
Why 88
Security100
Provenance78
Maintenance100
Licence100
Adoption51
Files (9)
- SKILL.md
- references/diagnostics.md
- references/false-progress.md
- references/glue.md
- references/observability.md
- references/rl/art.md
- references/serving/vllm.md
- references/sft/tinker.md
- references/trace-schema.md