Loading ToolBench

Reading the current corpus…

xanther-context-engine MCP server · D · ToolBench

xanther-context-engine

github.com/Xanther-Ai/xanther-context-engine

Open-source context engine for coding agents — architecture-aware code intelligence with cross-session memory. Provides MCP tools for semantic search, architectural analysis, impact analysis, and code traceability via Neo4j knowledge graph.

2026-07-28+STDIO5 tools

51/100

Definition Quality54

Protocol Readiness35

Supportability55

Spec posture

Inferred effective spec 2026-07-28+·current baseline 2026-07-28

Static source inference · medium confidence

Current-spec patterns detected

Summary

Five tools with complete JSON schemas and basic descriptions, but significant quality gaps. All tools have descriptions (10-60 chars, below the 50-200 char baseline for LLM optimization). Schemas present but lack depth: no output schemas documented, no pagination support despite search/index operations, no error recovery guidance. Parameter descriptions are minimal (e.g., 'Repository identifier' for repo_id). No tool annotations (readOnlyHint/destructiveHint). xce_index_repo is marked WRITE but has no confirmation/dry-run pattern. Missing idempotency guarantees and chaining field documentation. Naming is verb-forward (good), but descriptions do not explain WHEN to use each tool or what distinguishes them.

Tools (5)

xce_architecture_contextread onlyauthsource verified55/100

Get architectural context for a file or symbol

xce_impact_analysisread onlyauthsource verified53/100

Predict blast radius for proposed code changes

xce_index_repowriteauthsource verified48/100

Index or re-index a code repository

xce_searchread onlyauthsource verified55/100

Search the knowledge graph by semantic meaning or symbol

xce_traceread onlyauthsource verified57/100

Trace relationships between code and design artifacts

Issues (5)

CRITICALtool ↗

No output schemas documented. LLMs cannot plan downstream calls or extract required fields (e.g., does xce_architecture_context return file_path, symbol_name, dependencies?). Breaks tool chaining.

xce_architecture_contextxce_tracexce_impact_analysisxce_searchxce_index_repo

HIGHtool-description ↗

Descriptions are 10-60 characters (baseline 50-200 for LLM optimization). 'Get architectural context for a file or symbol' lacks WHEN to use it, what it returns, and how it differs from xce_search. Insufficient for LLM tool selection.

xce_architecture_contextxce_tracexce_impact_analysisxce_searchxce_index_repo

HIGHconfirmation-request ↗

xce_index_repo is marked WRITE (destructive) but has no dry-run, confirmation, or idempotency guarantee. Agents cannot safely retry on failure without risking duplicate indexing or data loss.

xce_index_repo

Recommendations

Expand all tool descriptions to 50-200 characters. Include WHAT the tool does, WHEN to use it (vs similar tools), and what it returns. Example: 'xce_architecture_context: Retrieve architectural layers, dependencies, and design patterns for a file or symbol. Use this to understand code structure before making changes. Returns: layers (list), dependencies (list), patterns (list).'
Document output schemas for all tools. Define return types as JSON Schema objects with typed fields. Example for xce_search: {type: 'object', properties: {results: {type: 'array', items: {type: 'object', properties: {symbol: {type: 'string'}, file: {type: 'string'}, score: {type: 'number'}}}}}, total: {type: 'integer'}}.
Add pagination to xce_search and xce_index_repo. Include limit (default 20, max 100), offset, and total_count in responses. Document: 'Results capped at 100 items per call. Use offset to paginate.'
Add tool annotations: readOnlyHint=true for xce_architecture_context, xce_trace, xce_impact_analysis, xce_search; destructiveHint=true for xce_index_repo. Include idempotentHint=true if xce_index_repo is idempotent (re-indexing same files is safe).
For xce_index_repo, add a dry_run parameter (boolean, default false) or require explicit confirmation. Document: 'Set dry_run=true to preview changes without modifying the index. Indexing is [idempotent/non-idempotent, clarify].' Add error recovery guidance: 'If indexing fails, call xce_index_repo again with incremental=true to resume.'

Score history

Overall score trend

First recorded score · v2 rubric

51/100

Scored	Grade	Overall	Spec posture	Rubric
2026-09-23	D	51	2026-07-28+	v2