Skill atlas
Skill luongs3/skill-atlas
A trust-rated index of public AI-agent skills, organized by job. Tells your agent which skills to load for a task — and whether to trust them (source, reputation, freshness).
npx -y skills add luongs3/skill-atlasAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Find the right public AI-agent skill for a job — and know whether to trust it. Load when about to start a task type (Upwork freelancing, technical interviews, office documents, MCP/tool building, prompt engineering, web/frontend, data analysis, learning English) and you want to know which existing public skills to pull in, rated by source reputation and freshness. Answers "which skill do I load for X, and can I trust it?"
SKILL.md
4.8 KB, ~1.2k tokens by cl100k_base, as published. Nobody here has run it
Skill Atlas
A trust-rated index of public AI-agent skills, organized by job. Use it to answer: "I'm about to do X — which public skills should I load, and can I trust them?"
The skills themselves live in their original repos. This atlas catalogs which ones are good, where they come from, and how stale they are — the trust layer most skill lists skip.
When to load this skill
At the start of a task that matches a known job, before hunting for tools yourself:
| If the task is about… | Read |
|---|---|
| Winning work on Upwork | jobs/upwork.md |
| Technical interview prep | jobs/interview-prep.md |
| Algorithms / system design | jobs/algorithms-system-design.md |
| Career planning / what to learn | jobs/career-roadmaps.md |
| Word/PDF/PowerPoint/Excel | jobs/office-documents.md |
| Building an MCP server / agent tool | jobs/mcp-and-agent-tools.md |
| Writing better prompts | jobs/prompt-engineering.md |
| Frontend / web app build or test | jobs/web-frontend.md |
| Writing idiomatic Go | jobs/go-development.md |
| Choosing Go backend libraries (HTTP, CLI, log, test) | jobs/go-backend-libraries.md |
| Git: rebase, recovery, history | jobs/git-version-control.md |
| Terraform / Kubernetes / infra | jobs/devops-infrastructure.md |
| Dockerfile / compose / containers | jobs/docker-containers.md |
| Schema design, SQL, migrations (Postgres) | jobs/databases-sql.md |
| Designing a REST or gRPC API | jobs/api-design.md |
| Metrics / traces / logs / dashboards | jobs/observability-monitoring.md |
| Caching, queues, scaling, Kafka/Redis | jobs/scalability-distributed-systems.md |
| Async messaging, task queue, workflows | jobs/message-queues-streaming.md |
| Applying a design pattern / refactor | jobs/software-design-patterns.md |
| AWS or GCP — CLI, SDK, services | jobs/cloud-aws-gcp.md |
| CI/CD — GitHub Actions, GitOps deploy | jobs/cicd-pipelines.md |
| Login, OAuth/OIDC, JWT, access control | jobs/authentication-authorization.md |
| Building a UI (React/Vue/Svelte/Next) | jobs/frontend-frameworks.md |
| Writing Go tests, mocks, integration | jobs/go-testing.md |
| Securing an app / authorized pentest | jobs/security.md |
| Exploring / cleaning / charting data | jobs/data-analysis.md |
| Social media / X research | jobs/social-media-research.md |
| Data pipelines, ETL, Airflow/Spark/dbt | jobs/data-engineering.md |
| Building a mobile app (Flutter/RN/iOS) | jobs/mobile-development.md |
| Building an app on LLMs (RAG, agents) | jobs/llm-app-development.md |
| Serving/running LLMs (Ollama/vLLM) | jobs/llm-serving-inference.md |
| Bash scripting, CLI, shell debugging | jobs/linux-shell.md |
| Configuring nginx / reverse proxy / TLS | jobs/nginx-web-servers.md |
| Finding a course/book to learn something | jobs/learning-resources.md |
| Improving English for work | jobs/learning-english.md |
How to use an entry
Each entry carries a trust tier plus source URL, reputation signal, and a
last_validated date. Workflow:
- Read the ranked list for your job. Prefer 🟢 A (canonical) and 🔵 B (community-proven).
- Treat 🟡 C as "read before trusting" and 🔴 D as "stale — caution."
- Load the public skill, then fork it private and adapt it to yourself. The public skill is the starting point; your private version encodes your voice, creds, and rules.
Trust tiers
- 🟢 A — Canonical: official vendor source (Anthropic, the spec author). Trust by authorship.
- 🔵 B — Community-proven: high reputation (stars/installs) and actively maintained.
- 🟡 C — Useful, verify: plausible but low/unknown reputation. Read before trusting.
- 🔴 D — Caution: stale (>12mo), unmaintained, or known-broken. Listed so you don't rediscover it.
Keeping it honest
scripts/revalidate.sh re-checks every source's liveness + GitHub stars/last-push (uses
gh for 5000 req/hr if available). An entry whose last_validated is >6 months old is
treated as C until reproven. A monthly GitHub Action runs this automatically and opens a
tracking issue if any source dies or goes stale — so the atlas can't silently rot. See
_meta/SCHEMA.md for the full schema and CONTRIBUTING.md for the trust bar on new entries.
Index repo: https://github.com/luongs3/skill-atlas — skills remain property of their original authors under their own licenses; this repo provides only the trust assessment.
Gives 0 of the 12 instructions most prompt engineering skills give in ~1.2k tokens
Counted across 563 of the 626 authors here whose files we hold, read 2026-08-06
- ask at most three clarifying questionsin 22 of 563, across 15 files
- respond in the user input languagein 14 of 563, across 9 files
- preserve the original intentin 13 of 563, across 11 files
- Establish baseline metrics and collect representative examplesin 12 of 563, across 2 files
- Identify failure modes and prioritize high-impact fixesin 12 of 563, across 2 files
- Apply prompt and workflow improvements with measurable goalsin 12 of 563, across 2 files
- Roll back quickly if quality or safety metrics regressin 12 of 563, across 2 files
- validate changes with tests and roll out in controlled stagesin 12 of 563, across 2 files
- generate quantitative baseline performance reportsin 12 of 563, across 2 files
- create representative test scenariosin 12 of 563, across 2 files
- treat prompts as codein 12 of 563, across 5 files
- test prompts on diverse inputsin 12 of 563, across 8 files
Said here and by no other author read
- Load this skill at the start of a task
- Read the specific file for your job
- Prefer canonical or community-proven skills
- Verify useful but unproven skills before trusting
- Treat stale skills with caution
- Fork public skills privately
Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.