OneWave AI Claude Skills
A library of 229 skills for work and life: sales, marketing, finance, engineering, design and parallel agent runs
- 368 ★ on the repository
- Last commit
- MIT
Broad access
Why this level
- The skills handle sensitive money and data: statements, invoices, contracts, CRM records, and with connected mail they read emails and create drafts
- The `agent-army` skill launches dozens of parallel agents that edit files in bulk and use up model limits
- Some skills run Python scripts on the user's files and reach the network, for example for SEC EDGAR filings
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/onewave-ai-claude-skillsDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- The skills handle sensitive money and data: statements, invoices, contracts, CRM records, and with connected mail they read emails and create drafts
- The `agent-army` skill launches dozens of parallel agents that edit files in bulk and use up model limits
- Some skills run Python scripts on the user's files and reach the network, for example for SEC EDGAR filings
This is third-party code. Review the repository files before installing.
What it does
A library from the consulting company OneWave AI. By the authors' count it holds 229 skills in three areas: business (sales, marketing, consulting, operations), everyday life (personal finance, travel, fitness, job hunting) and coding (engineering, design, AI agent architecture). Each skill sits in its own folder as a single `SKILL.md` file with `name` and `description` fields: the agent uses the description to decide when to load the skill, and the optional `tools` and `model` fields restrict the tools and pin the model. The agent architecture section gathers `agent-army`, `agent-to-agent`, `sub-agent-orchestrator` and `mcp-server-builder`. The flagship `agent-army` launches 3 to 50 or more independent agents, each with its own sub-agents, and works in waves: execute, audit, propagate. You set a power level that decides the mix of Opus, Sonnet and Haiku models in each layer. The `scout` skill recommends which skill fits a task, and `scout-pro` designs chains of several skills. Skills in the Work Tools section ship with Python scripts and reference data: `bookkeeping-close` reconciles a bank statement against the ledger, `spreadsheet-qa` answers questions about spreadsheets by computing in code and shows the rows and filters it used, and `sec-filing-puller` takes figures from 10-K and 10-Q filings on SEC EDGAR with a citation for each. Skills with the `cowork-` prefix are built for Claude Cowork and a connected mailbox: for example, `cowork-inbox-triage` sorts incoming mail and prepares reply drafts, and only a person sends the emails.
Who it is for. For founders, sales and marketing staff, consultants and developers who want to give an agent ready workflows from a single library.
Good fit when
- You need a ready skill for a typical business task: a proposal, a sales forecast, a contract review, a hiring plan
- A table from a CRM, Stripe or QuickBooks has to be checked by code rather than by eye: `spreadsheet-qa` fits
- A large edit across dozens of files should be split among parallel agents with git checkpoints: `agent-army` is there for that
- You are unsure which skill fits the task and ask the agent to suggest one through `scout`
- You need skills for personal tasks: a budget, a travel route, a workout program
Not a fit when
- You need a specialized tool for one field, such as accounting: the library skills are general-purpose and mostly consist of a single instruction file
- The edit is small, under six independent parts: the author of `agent-army` himself advises working directly in such cases
- Your tasks depend on Russian services and documents: the examples are built on QuickBooks, Stripe, Shopify, SEC EDGAR and US tax practice
- You need a single authorial voice across the whole result: by the author's account, parallel agents drift in voice
- You want the agent to send emails for you: the mail skills prepare drafts and leave sending to a person
Example request
Use the spreadsheet-qa skill and work out from our Stripe orders export what revenue was by month and how many customers returnedLimitations
- The counts disagree: the repository description says 200 or more skills, the description file says 229, and the tree holds 232 `SKILL.md` files because three skills are duplicated in the `plugins` folder.
- Most skills are a single file with no dependencies. Python scripts and tests sit with some skills, for example `bookkeeping-close`, `spreadsheet-qa` and `sec-filing-puller`, and they need Python installed.
- Skills with the `cowork-` prefix are built for Claude Cowork and connected Microsoft 365 or Gmail mail and calendar tools.
- The `contact-hunter` skill collects contacts of people and companies from open sources: use it only on data you are entitled to collect and process under the law.
- The repository is run by a consulting company, and its description links to the company's services and training. Skills that need their own manifest and build step were moved by the author to the neighboring Open Agent Stack repository. MIT license.
How to disable. Delete the folder of the skill from your agent's skills folder, for example `~/.claude/skills`, and if you cloned the whole library, delete the cloned folder. Files and drafts that skills already created stay where they are.
FAQ
How do I install one skill rather than the whole library?
In Claude Code with `claude skill install OneWave-AI/claude-skills/<skill-name>`, where `<skill-name>` is the folder name, for example `spreadsheet-qa`. For Codex CLI the author suggests cloning the repository into `~/.codex/skills`, with the same files unchanged.
Which agents do the skills work with?
By the author's description the skills follow the open Agent Skills standard, so the same files work in Claude Code, Codex CLI, Gemini CLI and Cursor: point the tool's skills folder at the clone. The description gives no test results for each of the agents.
What is agent-army and how many agents does it launch?
A two-layer structure of parallel agents. The first layer has 3 to 50 or more specialists, and each has at least two helpers on the second. The commander runs the work in waves, makes git checkpoints, checks the work before launch and catches cases where an agent reported a task as done without doing it. It suits tasks of six or more independent parts.
How do Work Tools skills differ from the rest?
They come with Python scripts and reference data, not only instruction text. For example, the `spreadsheet-model-auditor` script checks an Excel workbook for hardcoded constants, broken ranges, circular references and totals that do not foot. `crm-data-cleanup` has deduplication and evaluation scripts.
Do the skills send emails by themselves?
By its own rules `cowork-inbox-triage` creates drafts only, and a person sends the emails. It also does not archive emails that mention money, contracts or legal matters.
What are the three plugins in the plugins folder?
`agent-army`, `scout` and `scout-pro`. They have a `marketplace.json` manifest with the set name `onewave-skills` and a description of each plugin. The other skills are not part of that manifest.
Discussion
No questions or reviews yet.
Related
SkillsClaude Skills by borghei
Claude Skills
An installable skill library for teams: engineering, product, marketing, compliance, finance, with ready Python tools and agents
Claude Skills (Alireza Rezvani)
Claude Code Skills & Plugins
A large library of skills, agents and commands by domain: engineering, marketing, product, compliance, C-level advisors and research
A set of 13 SKILL.md skills: research, a product doc, decks, articles and audits for any compatible agent
A catalog of 28 mental-model skills for agents: diagnosis, decisions, systems, risk, strategy and idea generation
A personal set of six skills: goal briefs for an agent, disk cleanup, AI news digests, project tidy-up, deep research and long-form writing
Erich Owens' large Claude Code skill marketplace: design, computer vision, engineering, career, plus two MCP servers
Comparison with similar tools
Entries that share the main search query, and how they differ.
| Entry | Official | Repository stars | Main difference |
|---|---|---|---|
| OneWave AI Claude Skills (this entry) | no | 368 | |
| Claude Skills by borghei | no | 969 | a library of skills for teams: engineering, product, marketing, compliance, finance, with ready Python tools and agents |
| Claude Skills (Alireza Rezvani) | no | 28,576 | a large library of skills, agents and commands by domain: engineering, marketing, product, compliance, C-level advisors and research |
README in short
The description opens with the three areas and a link to the neighboring Open Agent Stack repository with 27 multi-file artifacts. Then come a block about `agent-army`, three install commands, a table of contents of fourteen sections with skill tables, and a description of the `SKILL.md` format with its required and optional fields. The end covers the authors, links to their services and rules for those who submit new skills.
README badge
Original SKILL.md
--- name: agent-army description: "Deploy a 2-layer parallel agent hierarchy for large, parallelizable work — big refactors, multi-file migrations, codebase-wide audits, bulk generation. A top-tier commander (Fable or Opus) orchestrates the swarms; the user picks a power level (Max Power / Heavy / Balanced / Economy) that sets the Opus/Sonnet/Haiku model mix per layer. Layer 1 is 3-50+ specialist agents, each with its own full context window; Layer 2 is 2+ sub-agents per member. Includes git safety, tiered sizing, a pre-deploy gate, phantom-completion checks, and multi-wave follow-up." user_invocable: true --- # Agent Army A 2-layer parallel execution framework. The Commander is the top-tier model — Fable by default, or Opus if that's the session model — and does the thinking: recon, composition, briefing, verification. It orchestrates swarms of subordinate models whose tiers the user picks at deploy time (see Power Level). Each Layer 1 agent has its own full context window (not a slice). Each spawns Layer 2 sub-agents under it. The result is many independent brains running at once — not one brain divided. ``` Commander (you — Fable or Opus, the session model) | |-- Layer 1: Team (3 to 50+, each = own 1M context, L1 tier from power level) | |-- Agent A (1M) -- Sub-agent A1, A2, ... (L2 tier) | |-- Agent B (1M) -- Sub-agent B1, B2, ... (L2 tier) | |-- Agent C (1M) -- Sub-agent C1, C2, ... (L2 tier) | |-- ... (no cap) ``` **Swarm vs. army:** A swarm splits one context window across sub-agents — one brain, divided. An army gives each Layer 1 member its own window. That difference is the whole point of this skill. ## When to use - Large refactors spanning many files - Multi-file color / style / naming / API migrations - Broad codebase audits (security, a11y, performance, dead code) - Bulk content generation or transformation - Any task with **6+ independent units of work** that can run simultaneously ## When NOT to use - Fewer than 6 independent units of work — just do it directly - Heavy sequential dependencies where each step needs the last one's output - Single-file changes - Tasks needing one coherent authorial voice across all output (parallel agents drift) If the task doesn't clearly fit, say so and propose doing it inline instead of spinning up an army. <mandatory-rules>