Runtime-agnostic Generative Engine Optimization (GEO) toolkit and CLI for analyzing, ranking, rewriting, and schema-marking web pages to improve citation by answer engines (ChatGPT, Perplexity, Gemini, Claude).
E-GEO provides 6 tools with descriptions and basic parameter schemas, but exhibits significant gaps in parameter documentation, output schema clarity, and error handling guidance. Tool names are action-oriented (analyze, rank, rewrite, index, collect_page, collect_serp), which is positive. However, most parameters lack type definitions in visible schemas, descriptions are present but often generic, and output structures are not formally documented. The server is designed as a CLI/STDIO tool for GEO analysis, not as a remote MCP server, limiting its applicability to the MCP ecosystem.
Score content against the 10 GEO features (ranking_emphasis, user_intent, competitive_diff, social_proof, narrative, authority, usps, urgency, scannable, factual) with a transparent heuristic. Deterministic, offline, no LLM required.
Page collector: deterministic, zero LLM calls. Extracts title, meta description, JSON-LD @types, visible text, and content hash from a URL. One record per URL per pass appended to $EGEO_HOME/data/page/<page-slug>.jsonl.
SERP collector: position history for queries from Brave Search API or SerpBase Google Search API. Deterministic, zero LLM calls. One record per query per pass appended to $EGEO_HOME/data/serp/<query-slug>.jsonl.
Deterministic JSON-LD template filler. Generates a JSON-LD document for a given schema type (Article, Organization, Product, Service, FAQPage) populated with content metadata.
Simulate AI-engine ranking by delegating to geo_eval._rank_candidates. Orders candidate descriptions by relevance to a query using an LLM.
Optimize a description by delegating to geo_eval._rewrite_description. Rewrites a meta description to improve GEO signals and answer-engine citation likelihood.
STDIO-only transport: server is not remotely accessible and cannot integrate with hosted MCP clients. Limits deployment to local CLI use only.
Output schemas not documented: tools return results but no formal schema is visible for what fields/types agents should expect. Rank, rewrite, and analyze lack documented return structures.
Parameter type definitions incomplete: schemas show parameter names and descriptions but lack explicit JSON Schema type declarations (string, integer, array, object). Collect_page and collect_serp have 'budget' as integer but others lack type clarity.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 51 | 2026-07-28+ | v2 |
Error handling guidance absent: no documented error responses, recovery paths, or actionable error messages. Tools delegate to external modules (geo_eval._rank_candidates, geo_eval._rewrite_description) but failure modes are not described.
No tool annotations: tools lack readOnlyHint, destructiveHint, or idempotentHint metadata. Collect_page and collect_serp append to JSONL files (stateful side effects) but are not marked as destructive or non-idempotent.