Home

OmniRank

OmniRank audits a website's SEO, AEO, GEO, performance and security signals against its real HTML, generates llms.txt, llms-full.txt and facts.json for citation, and locates the exact source file behind each finding. Version 0.4.0 ships two Claude Code skills, audit and geo-artifacts, plus the omnirank fix diff preview, which writes nothing.

Three audiences — search crawlers, answer engines, and generative AI systems — parse the same rendered HTML. A page that is crawlable, has a liftable answer near the top, and publishes machine-readable ground truth wins all three at once, which is why OmniRank audits one artifact instead of running three separate tools. Since v0.3.0 it can also say which file is wrong, via the locator described in The-Locator. v0.4.0's theme is "OmniRank finds where your site contradicts itself" — five new gates catch a site's own declarations disagreeing with each other, at 100% precision, plus a new security layer and Google rich-result validation. See Contradictions and Security-Layer.

What ships in v0.4.0?

Skill	Status	What it does
audit	Shipped	Scores SEO, AEO, GEO, performance and security gates against a site's real HTML, including a site-level cross-URL pass and an indexability-contradictions pass, and reports observed / expected / fix for every gap. See Audit-Skill.
geo-artifacts	Shipped	Generates llms.txt, llms-full.txt and facts.json, each with an explicit citation licence. See GEO-Artifacts-Skill.
omnirank fix (CLI preview)	Shipped	Resolves a finding's URL to its source file and prints the diff it would apply for the four mechanical findings. Writes nothing — there is no --write flag. See Fix-Preview.

That is the entire shipped surface of v0.4.0. fix is a CLI subcommand of the audit skill's package, not a third registered Claude Code skill — .claude-plugin/plugin.json still declares exactly two skills. No package is published to PyPI, npm, or a container registry, and no benchmarks, user counts, or testimonials exist for this project — everything on this wiki is either transcribed from source or explicitly marked as unverified.

What's new in v0.4.0

• A security layer — five content layers now, not four: SEO, AEO, GEO, perf and security. Eight new finding ids; four response-header gates are inventory-only and never graded, two (mixed-content, https-redirect) are graded because they break crawling or rendering. See Security-Layer.
• Five indexability-contradiction gates — a sitemap URL its own robots.txt disallows, a canonical pointing at a noindexed/missing/redirecting page, an hreflang alternate that is itself noindexed. 100% precision by construction. See Contradictions.
• Google rich-result required-property validation — seo.schema-required.missing-property checks Google's documented requirements, distinct from schema.org validity.
• Four on-page accessibility-overlap gates — missing alt, skipped heading levels, empty/generic link text, missing lang.
• A corrected scoring model — each layer's budget now scales with its own gate count instead of a flat 100-point constant. Scores from v0.3.0 and v0.4.0 are not comparable. See Audit-Skill#how-is-the-score-computed.

67 finding ids total (up from 48), across 42 --fail-on gate names (up from 28).

What's on the roadmap?

Skill	Target	What it will do
aeo-onpage	v0.2 (slipped)	Emit JSON-LD by entity type; draft AnswerBlocks and FAQs; Next.js codegen
indexing	v0.3	IndexNow, GSC URL Inspection, Bing Submit, Wayback, hash-based freshness
offsite-entity	v0.4	sameAs gap analysis, peer mention-gap detection, outreach drafts for human review
measure	v0.5	Rank tracking plus real AI-citation testing across engines
smm-content	v0.6	Repurpose published pages into platform-native assets
smm-publish	v0.7	Gated publishing — dry-run by default, human approval required

Two targets have already slipped. aeo-onpage's "v0.2" predates this table: 0.2.0 shipped as a site-gates/AnswerBlock-bands/perf release instead, folded into the existing audit skill. offsite-entity's "v0.4" also slipped: v0.4.0 shipped, and it spent its budget on the security layer, the contradiction gates, and the scoring fix instead — nothing about offsite-entity is built. Treat every target version as directional, not a commitment. None of these six skills exist in the installed package today — asking Claude Code to "write our JSON-LD" or "submit this URL to Google" will not trigger anything. Full detail: Roadmap.

How do I run my first audit?

Three commands, no configuration file required:

git clone https://github.com/bemoshiur/OmniRank.git && cd OmniRank
python3 -m venv .venv && source .venv/bin/activate && make install
python3 -m omnirank.cli audit https://example.com

To see what OmniRank would fix, without writing anything:

python3 -m omnirank.cli fix https://example.com --root .

Full walkthrough with real pasted output, exit codes, and what to do if a step fails: Quick-Start.

Where do I go next?

Page	What it covers
Quick-Start	Zero to first audit, numbered steps, real terminal output
Configuration-Reference	Every omnirank.config.json field, generated from the JSON Schema
Audit-Skill	The audit skill and CLI: every gate, severity, trigger and fix; the scoring formula
Security-Layer	The new security layer: what each gate detects, and its honest limits
Contradictions	The new indexability-contradiction gates: a real true positive, and their limits
GEO-Artifacts-Skill	llms.txt / llms-full.txt / facts.json; Python vs Node paths; the OpenNext 403 trap
Fix-Preview	What omnirank fix does, its flags, and why it writes nothing
Fix-Tiers-and-Applicability	The fixTier × applicability model that decides what is safe to fix
The-Locator	How OmniRank resolves a URL to the source file that owns it
Finding-Reference	All 67 finding ids, generated from the registry — layer, gate, severity, fixTier
Report-Schema	The finding shape, the <layer>.<gate>.<condition> id convention, notEvaluated
CI-Recipes	GitHub Actions, GitLab CI, generic shell, and choosing --fail-on gates
Claude-Code-Setup	Installing via the release ZIP or a clone; real trigger phrases per skill
Troubleshooting	Real error text for likely failures, with the fix for each
FAQ	Direct, honest answers — including what OmniRank does not do
Glossary	SEO/AEO/GEO/SMM terms, each defined in a standalone, quotable answer
Research-and-Evidence	The published research this project builds on, cited precisely
Roadmap	Shipped vs planned, by target version

What OmniRank does not do

• It does not force rankings. Nothing can. It scores the signals search and answer engines are known to use and hands you a prioritised list of gaps to close.
• It does not send cold email. Outreach text is drafted for human review and manual send, never dispatched automatically.
• It does not generate AI slop. Every generated page expects a human edit pass before publishing.
• It does not edit your site. audit only diagnoses; geo-artifacts writes only the three GEO artifacts it generates; omnirank fix prints a diff and writes nothing — there is no --write flag. Writing ships once the locator is proven against real repositories and the write guarantees it depends on are implemented and tested, not on a release number.

See the project README's "What OmniRank does not do" section for the complete list, and FAQ for the honest, uncomfortable questions answered directly.

Project links

Source repository	github.com/bemoshiur/OmniRank
Documentation (in-repo)	docs/
Issues	github.com/bemoshiur/OmniRank/issues
Maintainer	S M Moshiur Rahman, TICON System Limited, Dhaka
Licence	Code MIT · Content CC BY 4.0