Master Skill Library - Router
A local, offline library of 35,802 reusable agent assets from 186 repositories.
Nothing here loads into context automatically. You search the catalog, then
read only the one or two files you actually need.
Library root
The library root is wherever you cloned/copied this repo
(the-mega-skill-library, github.com/thedixitjain/the-mega-skill-library).
All paths below are relative to that root — resolve them from CATALOG.tsv's
own location, not from wherever this router skill happens to be installed.
Step 1 - search the catalog
CATALOG.tsv is tab-separated, one row per asset:
kind <TAB> category <TAB> name <TAB> path <TAB> source_repo <TAB> description
Use the Grep tool on CATALOG.tsv with the user's task keywords. Shell equivalents:
grep -i "<keyword>" CATALOG.tsv | head -30 # anywhere
grep -i "<keyword>" CATALOG.tsv | awk -F'\t' '$1=="skill"' # only Agent Skills
awk -F'\t' '$2=="security-and-compliance"' CATALOG.tsv # whole category
Step 2 - read the matching asset
Column 4 is the path relative to this library root. Read that file.
For an Agent Skill, sibling files in the same directory (scripts, references,
templates) are part of the bundle and may be read or executed.
Step 3 - apply it
Follow the loaded file's instructions as if the skill had been invoked directly.
When several candidates match, prefer the one whose source_repo is an official
vendor: anthropics/, openai/, google/, microsoft/, github/,
vercel-labs/, android/, trailofbits/, modelcontextprotocol/.
Asset kinds
| Kind |
Path pattern |
What it is |
skill |
library/<category>/<slug>/SKILL.md |
Full Agent Skill bundle (18,681) |
agent |
agents/<category>/<slug>.md |
Subagent definition (2,888) |
command |
commands/<category>/<slug>.md |
Slash-command prompt (2,773) |
prompt |
prompts/<category>/<slug>.md |
System prompt / prompt corpus (10,074) |
rules |
rules/<category>/<slug>.md |
CLAUDE.md / AGENTS.md / .cursorrules (1,386) |
Plus reference/<owner~repo>/... - 543 verbatim awesome-lists,
cheatsheets and handbooks, greppable but not skill-shaped.
Categories
general-purpose — 5,575 — Uncategorized / cross-cuttingai-agents-and-harness — 4,583 — Agent design, orchestration, subagents, harness tuning, context engineeringprompt-engineering — 2,868 — Prompt techniques, system prompts, optimizers, persona designengineering-core — 2,805 — Refactoring, debugging, review, architecture, git, spec-driven dev, TDDresearch-and-academic — 2,483 — Literature review, papers, journals, grants, regulatory sciencedevops-and-infra — 1,605 — Docker, K8s, Terraform, CI/CD, cloud, observability, SREsecurity-and-compliance — 1,604 — AppSec, pentest, OSINT, threat modeling, SOC2/ISO/GDPR compliancemarketing-and-growth — 1,584 — SEO/AEO, ads, CRO, social, campaigns, growth loopsbackend-and-data — 1,583 — APIs, databases, schemas, queues, caching, data pipelinesmcp-and-integrations — 1,524 — MCP servers/clients, third-party integrations, browser automationtesting-and-qa — 1,506 — Test generation, E2E, coverage, QA process, browser testingfrontend-and-design — 1,359 — UI/UX, design systems, CSS, React, accessibility, visual polishwriting-and-content — 1,345 — Copywriting, editing, humanizing, translation, narrativedocs-and-knowledge-mgmt — 1,158 — Documentation, ADRs, technical writing, Obsidian, note systemsbusiness-and-finance — 1,152 — Finance, pricing, legal, sales, HR, exec/C-suite advisorymedia-and-creative — 778 — Video, image, slides/PPTX, audio, game dev, generative artdata-science-and-ml — 722 — ML/DL, training, fine-tuning, evaluation, statistics, notebooksproduct-and-pm — 473 — PRDs, roadmaps, user stories, discovery, product analyticsmobile-and-platform — 398 — iOS/Android/Flutter, desktop, extensions, Office/BI platformsproductivity-and-workflow — 353 — Planning, automation, meetings, organization, scaffolds, coachingrag-memory-knowledge — 333 — RAG, vector search, embeddings, agent memory, knowledge graphsskill-authoring-meta — 11 — Creating, testing, auditing and improving skills themselves
Provenance
Every generated file carries source_repo, source_path and source_url in its
frontmatter plus a Source footer identifying which upstream repo it came from.
1---2name: master-skill-library3description: Router for a local library of 35,802 agent skills, subagents, slash commands, system prompts and rules files harvested from 186 top GitHub repositories. Use whenever a task might already have a purpose-built skill - coding, review, testing, security, devops, design, data, RAG, writing, docs, product, marketing, business, research, media, mobile or productivity. Search CATALOG.tsv first, then read only the matching file.4---56# Master Skill Library - Router78A local, offline library of **35,802 reusable agent assets** from **186 repositories**.910Nothing here loads into context automatically. You **search the catalog**, then11**read only the one or two files** you actually need.1213## Library root1415The library root is wherever you cloned/copied this repo16(`the-mega-skill-library`, github.com/thedixitjain/the-mega-skill-library).17All paths below are relative to that root — resolve them from `CATALOG.tsv`'s18own location, not from wherever this router skill happens to be installed.1920## Step 1 - search the catalog2122`CATALOG.tsv` is tab-separated, one row per asset:2324```25kind <TAB> category <TAB> name <TAB> path <TAB> source_repo <TAB> description26```2728Use the Grep tool on `CATALOG.tsv` with the user's task keywords. Shell equivalents:2930```bash31grep -i "<keyword>" CATALOG.tsv | head -30 # anywhere32grep -i "<keyword>" CATALOG.tsv | awk -F'\t' '$1=="skill"' # only Agent Skills33awk -F'\t' '$2=="security-and-compliance"' CATALOG.tsv # whole category34```3536## Step 2 - read the matching asset3738Column 4 is the path relative to this library root. Read that file.39For an Agent Skill, sibling files in the same directory (scripts, references,40templates) are part of the bundle and may be read or executed.4142## Step 3 - apply it4344Follow the loaded file's instructions as if the skill had been invoked directly.45When several candidates match, prefer the one whose `source_repo` is an official46vendor: `anthropics/`, `openai/`, `google/`, `microsoft/`, `github/`,47`vercel-labs/`, `android/`, `trailofbits/`, `modelcontextprotocol/`.4849## Asset kinds5051| Kind | Path pattern | What it is |52|---|---|---|53| `skill` | `library/<category>/<slug>/SKILL.md` | Full Agent Skill bundle (18,681) |54| `agent` | `agents/<category>/<slug>.md` | Subagent definition (2,888) |55| `command` | `commands/<category>/<slug>.md` | Slash-command prompt (2,773) |56| `prompt` | `prompts/<category>/<slug>.md` | System prompt / prompt corpus (10,074) |57| `rules` | `rules/<category>/<slug>.md` | CLAUDE.md / AGENTS.md / .cursorrules (1,386) |5859Plus `reference/<owner~repo>/...` - 543 verbatim awesome-lists,60cheatsheets and handbooks, greppable but not skill-shaped.6162## Categories6364- **`general-purpose`** — 5,575 — Uncategorized / cross-cutting65- **`ai-agents-and-harness`** — 4,583 — Agent design, orchestration, subagents, harness tuning, context engineering66- **`prompt-engineering`** — 2,868 — Prompt techniques, system prompts, optimizers, persona design67- **`engineering-core`** — 2,805 — Refactoring, debugging, review, architecture, git, spec-driven dev, TDD68- **`research-and-academic`** — 2,483 — Literature review, papers, journals, grants, regulatory science69- **`devops-and-infra`** — 1,605 — Docker, K8s, Terraform, CI/CD, cloud, observability, SRE70- **`security-and-compliance`** — 1,604 — AppSec, pentest, OSINT, threat modeling, SOC2/ISO/GDPR compliance71- **`marketing-and-growth`** — 1,584 — SEO/AEO, ads, CRO, social, campaigns, growth loops72- **`backend-and-data`** — 1,583 — APIs, databases, schemas, queues, caching, data pipelines73- **`mcp-and-integrations`** — 1,524 — MCP servers/clients, third-party integrations, browser automation74- **`testing-and-qa`** — 1,506 — Test generation, E2E, coverage, QA process, browser testing75- **`frontend-and-design`** — 1,359 — UI/UX, design systems, CSS, React, accessibility, visual polish76- **`writing-and-content`** — 1,345 — Copywriting, editing, humanizing, translation, narrative77- **`docs-and-knowledge-mgmt`** — 1,158 — Documentation, ADRs, technical writing, Obsidian, note systems78- **`business-and-finance`** — 1,152 — Finance, pricing, legal, sales, HR, exec/C-suite advisory79- **`media-and-creative`** — 778 — Video, image, slides/PPTX, audio, game dev, generative art80- **`data-science-and-ml`** — 722 — ML/DL, training, fine-tuning, evaluation, statistics, notebooks81- **`product-and-pm`** — 473 — PRDs, roadmaps, user stories, discovery, product analytics82- **`mobile-and-platform`** — 398 — iOS/Android/Flutter, desktop, extensions, Office/BI platforms83- **`productivity-and-workflow`** — 353 — Planning, automation, meetings, organization, scaffolds, coaching84- **`rag-memory-knowledge`** — 333 — RAG, vector search, embeddings, agent memory, knowledge graphs85- **`skill-authoring-meta`** — 11 — Creating, testing, auditing and improving skills themselves8687## Provenance8889Every generated file carries `source_repo`, `source_path` and `source_url` in its90frontmatter plus a Source footer identifying which upstream repo it came from.
thedixitjain