Nine skills that make your AI agent ask the hard questions before you spend a year building: is it worth it, can AI agents use it, will it still matter in five years, what can you protect, and how do you sell it.
curl -fsSL https://eligapris.com/founder-skills.sh | shPaste it in a terminal. Read the installer first
Writing the code too? Get the Dev Skill Pack
How it works
It copies nine skills into ~/.agents/skills in a few seconds, for your user only.
Each skill is linked into every agent on your computer, and the writing rule goes into their global instructions so it's always on.
"Is this idea worth building?" "Is my app ready for agents?" "Will it last five years?" The agent picks the right skill by itself.
Why use it
Checks whether your idea already exists as a free AI skill or MCP server, runs the economics, then sets up a 7-day test of the problem in your own market, where global numbers can mislead.
Deep research with sources ranked by how reliable they are, saved to a local knowledge base that grows with every question.
Scores your product on 10 questions about whether AI agents can use it (API or MCP, CLI, keys, approvals, pricing), then converts it.
Forecasts five years ahead: can agents use your product, will better models make it unnecessary, and which move raises your odds most.
Finds what's actually new in your code or idea, and whether it's worth a patent, a copyright, a trademark, or keeping secret.
Positioning and messaging that starts with why anyone should care, before features.
Edward Bernays's Propaganda (1928) turned into playbooks for launches, narrative, and trust. The full book text is included.
Finds the weak spots in a landing page or sales copy and rewrites every line to convert.
Always on: your agent stops writing like a chatbot, in every pitch, email, and post. It comes with a checker for what slipped through.
Skills with the same name as yours are moved to ~/.agents/skill-backups first, and instruction files are backed up before the writing rule goes in.
Skills are instructions and a few helper scripts in a folder on your machine. The installer downloads one zip from eligapris.com, adds the writing rule to your agents' instruction files, and sends back one anonymous install report (pack, version, OS, which agents it set up). Run it with DO_NOT_TRACK=1 to skip that. What your agent researches goes through your own agent and its usual privacy terms.
Read exactly what we recordFAQ
No, but you need an AI agent that can run commands: Claude Code (including the Code tab in the Claude desktop app), Codex, Cursor, Gemini CLI, or OpenCode. Easiest way: paste the "Ask your agent" line into it and let it do the install.
On macOS or Linux, open a terminal and paste curl -fsSL https://eligapris.com/founder-skills.sh | sh. On Windows, open PowerShell and paste irm https://eligapris.com/founder-skills.ps1 | iex. Or paste "Set up the skill pack from https://eligapris.com/founder-skills.md" into your agent. Then start a new session.
lean-venture-analyst checks first. It searches skills.sh, SkillsMP, mdskills.ai and the official MCP registry for skills and MCP servers that do the same job, with their install counts. A free skill thousands of people use means your product has to sell something it can't, like the finished result, your data, trust, or a local presence. Skills that exist but nobody installs tell you the job is known and still unsolved for real users.
Because global data can be wrong for your area. A space can look crowded or dead worldwide while the city, country or customers you serve haven't adopted the technology yet, or look booming while your customers still solve it another way. The skill uses global research to plan the test, then lets 7 days of real conversations where you sell decide whether the problem is real there.
More and more people get work done by asking an agent instead of clicking through an app. A product is ready when an agent can do its main jobs: through an API or MCP server, with a key it can use without a browser, with docs it can read, and with a human approval step for anything risky. agent-led-product checks all ten of these and shows what's missing.
Your agent gathers the evidence (your code, competitors, what a plain AI agent can already do without your product) and a script does the math, so the same evidence always gives the same numbers. You get a Year 0 to Year 5 curve with best, expected, and worst cases, and the moves that raise Year 5 the most.
Yes. The two packs share two skills (agent-led-product and humanize) and install cleanly in either order; the writing rule is never added twice.
Yes. Open your agent's instructions file (for example ~/.claude/CLAUDE.md) and delete everything between the dev-skill-pack:humanize start and end markers.
One anonymous report when it finishes: which pack and version, your OS and processor type, which agents it set up, and whether it was the one-line or zip install. No paths, usernames, file contents or anything from your projects. The install counts on this page come from those reports. To skip it, put DO_NOT_TRACK=1 in front of sh, for example curl -fsSL ... | DO_NOT_TRACK=1 sh, or set it in PowerShell with $env:DO_NOT_TRACK=1 first.
Run the same line again to update. To remove it, delete the skill folders from ~/.agents/skills (and their links in your agents' skills folders), and the dev-skill-pack:humanize block from your agents' instruction files.
More free tools
Passport, visa, and upload photos
Photo Resizer
Pick where the photo is going (passport, visa, or a size limit) and get a file that passes, first time.
OpenStrip hidden data from files
Metadata Remover
Remove GPS location, camera details, and author names from photos, PDFs, and documents.
OpenSee what a file reveals
Metadata Viewer
See where a photo was taken, which device made it, who wrote a document, and whether it says it's AI.
OpenWant tools like this in your own product?
Eligapris builds fast, private web apps and on-device AI. The same engineering behind Founder Skill Pack, for your team.