Engineering
Design docs, reviews, incident write-ups, and release checks.
Yasir Jilani
A local library of 601 agent skills. Each one has a scenario, sample data, and a filled outcome. Built for Claude, and for any tool that reads the Agent Skills standard.
Each skill is a folder with a SKILL.md file. Claude reads the short description at startup and loads the procedure only when the task matches. Install one domain so you do not pay for descriptions you are not using.
python3 scripts/install.py --tool claude --domain finance python3 scripts/install.py --tool claude --domain finance --dry-run
Run it from the repository root. Finance skills land in ~/.claude/skills/<skill-name>/, not in a finance subfolder. The installer copies files on this machine. It does not download a script, ask for a key, or run the skills. Naming and the write format are in CONTRIBUTING.md.
Can we hire two engineers next month without missing payroll?
That is not a pep talk. It is a cash question. The skill cash-flow-forecast builds a 13-week view from collections and commitments, then shows the week the buffer breaks if a receipt slips.
The skill will not invent a collection rate, ask for a bank password, or call the result investment advice. Figures you did not supply stay labeled as assumptions.
See examples/by-skill/finance/cash-flow-forecast.md.
Design docs, reviews, incident write-ups, and release checks.
Positioning, campaigns, and claims you can actually support.
Cash, runway, margin, and management packs. Not investment advice.
Hiring, reviews, and pay bands without invented market data.
Choices, board memos, and an operating cadence.
Discovery, qualification, and proposals without fake proof.
Briefs, PRDs, experiments, and roadmaps with kill criteria.
Contract triage and counsel briefs. Not a legal opinion.
Close, reconciliations, and audit prep. Not an audit opinion.
Controls, access, and response. No exploits or bypasses.
SOPs, capacity, and service levels a team can staff.
Contracts, lineage, freshness, and readouts from files you have.
Also included: AI, data, SEO, blogs, news, SaaS, startups, small business, supply chain, oil and gas, energy, automotive, and airlines. Every skill has a scenario file under examples/by-skill/.
plugins/finance/skills/cash-flow-forecast/ ├── SKILL.md ├── references/assumptions.md └── scripts/cashflow_check.py
Frontmatter has a name, a description that says when to use the skill, the MIT license, and the author. The body is the procedure: inputs, steps, a quality bar, an example, and the anti-patterns. Long notes stay in references/ so they load only if needed.
| Check | What you should see |
|---|---|
| Installer | A local Python copy. No remote script. |
| Account | None. No license server, wallet, or activation key. |
| Scripts | Standard library only. They read files you pass in. |
| Integrity | MANIFEST.sha256 is verified before copy. |
| Affiliation | Independent work by Yasir Jilani. Not an Anthropic product. |
Skills draft work for a person to review. They do not replace counsel, a clinician, a CPA, a licensed broker, or a credit officer. If a request asks the agent to deceive someone or evade a control, the skill tells it to stop.
/plugin marketplace add /path/to/yj-claude-skills /plugin install finance@yj-skills
After you publish the folder to GitHub, teammates can add that repository as the marketplace. Until then, the path install and scripts/install.py are the supported paths. Both stay on the machine.