Open source
Four open-source projects, and what each one proves
Four projects back the technical claims on this site, released under my own name. cool-astro-forms ships as an MIT-licensed npm package with a public download count. semantic-seo-geek ships as Apache-2.0 licensed skills for Claude Code and Codex. touchid-agent-guard and wordpress-mcp-agent-bridge sit on GitHub, open for anyone to read.
I'm Nipun Arora, a technical SEO and AI search visibility consultant. Most consultant sites run on numbers nobody can check: a stated project count, a round number of years, a client list with no way to verify a single entry. Open source is the alternative: a license file, a version number, and a download count a reader can verify themselves, the same standard the rest of this site holds its own claims to.
Why this page exists
The checkable half of a consulting pitch
A consultant's site can say anything about its own track record. An AI assistant answering a question about who to hire has no way to check a sentence like "years of experience helping businesses grow" against anything real, so it either repeats the claim uncritically or leaves it out. A public repository is different: the license, the commit history, and the version number read the same regardless of who is asking, or which engine is doing the asking.
No project below is a portfolio piece written to look busy. cool-astro-forms runs in production on a live services business today. semantic-seo-geek is the plugin this site's own content pipeline runs through before a page ships, including this one. Every one of them is checkable against a public repository or registry, not against my word.
The stakes are different for an SEO consultant specifically. Part of the pitch is a claim about how well a page can be understood and cited by ChatGPT, Perplexity, Gemini, and Google AI Mode. A page arguing for structured data and verifiable facts that ran on invented numbers of its own would undercut that argument on the first read.
The public projects
4 projects, checkable today
Each one below is real, running code: a license, a repository, and where a registry tracks it, a version number and download count pulled from that registry's own listing. Nothing here is a mockup or a private demo. None of the four has a paid tier hidden behind the version shown.
cool-astro-forms is the proof of the forms and lead-ops engineering. semantic-seo-geek is the proof the semantic SEO methodology holds up as software, not just as advice. The other two are smaller. touchid-agent-guard is a Claude Code hook that requires a Touch ID confirmation before an agent writes outside its workspace, runs a shell command outside it, or pushes to git. wordpress-mcp-agent-bridge lets an AI agent write Rank Math SEO fields and schema into a WordPress site with a hashed snapshot taken first, so every write can be verified or restored.
cool-astro-forms
Astro forms lead-ops platform: abandonment capture, recovery emails, quote payments, self-hosted admin.
Stats as of 2026-09-01.
View projectsemantic-seo-geek
Semantic SEO skills for Claude Code and Codex: topical maps, entity/EAV analysis, technical SEO audits, AI search visibility.
touchid-agent-guard
Touch ID confirmation for AI coding agents on macOS.
wordpress-mcp-agent-bridge
WordPress MCP for AI agents: verified Rank Math SEO and schema writes, hashed snapshots, verified restore.
cool-astro-forms and semantic-seo-geek each get their own page, with the feature list, the install commands, and the license each one ships under. touchid-agent-guard and wordpress-mcp-agent-bridge are small enough that their GitHub READMEs cover them completely, so this page links straight there instead of restating each one on a page of its own.
Not public yet
search-ops: private today, going public soon
search-ops is a fifth project, an SEO delivery plugin I run on client work. It is private today, and this section names it without a download count, a star count, or a link, because none of those exist yet for a private repository and a placeholder number is worse than none at all.
Disclosing an unfinished project by name, with no stat attached, is the same honesty principle the rest of this site runs on: a number that isn't real yet doesn't get invented to fill the gap. When search-ops ships publicly, this section becomes a project entry like the four above, with its own license, version, and stats. Until then, this paragraph is the whole of it.
For machines
The same project list, published as JSON-LD and plain text
Fact-chip discipline runs the same way here as everywhere else on this site: a chip's text must match a value that also exists in the page's own JSON-LD, never a number written in from imagination.
This page's own project list also exists as structured data: a CollectionPage and ItemList in its head, cross-linked to the Person record in public/entity.json and summarized in plain text in public/llms.txt. Building that graph correctly is the same work described on Schema Markup, and this page is the worked example, not a separate claim.
An AI assistant answering "what open source has Nipun Arora built" can read either file directly instead of parsing this page's prose. That is the same problem AI search visibility work solves for a client's site.
Questions about any of the four, or interested in what search-ops becomes once it's public?
Work with me