ServiceNow Atlas — local-first AI knowledge layer for ServiceNow development via MCP. Provides filesystem navigation and semantic search over ServiceNow documentation and code examples.
servicenow-atlas provides 9 well-named, read-only tools with complete input schemas and reasonable descriptions. Naming follows verb_noun convention (list_*, search_*, read_*, get_*). All tools have descriptions (avg ~120 chars, within baseline 34-392). Input schemas are present with type definitions and parameter descriptions. However, output schemas are not documented in the source, LLMs cannot see what fields to expect from responses. Error handling is minimal; no recovery guidance or actionable error messages visible. No tool annotations (readOnlyHint, etc.) despite all being read-only. Parameter descriptions are adequate but lack format/constraint details (e.g., max_chars default is stated but no guidance on typical values). The server is STDIO-only, which caps protocol readiness at 50 regardless of definition quality.
Return metadata about the loaded bundle: model ID, embedding dimension, chunk count, publication list, and build timestamp.
Retrieve a single chunk by its ID. Useful for fetching full text after a search result.
Return the current docs release metadata: branch, commit SHA, last commit date, total file count.
List all files in a single publication, with title/product_area/last_updated from YAML frontmatter.
List every ServiceNow documentation publication (top-level folder) available in the local clone.
Read a single markdown file. Returns parsed YAML frontmatter and the body content.
Output schemas not documented. LLMs cannot see what fields responses contain (e.g., does search_docs return 'score' or 'similarity'? 'chunk_id' or 'id'?). This forces agents to guess field names for downstream tool calls.
No tool annotations despite all 9 tools being read-only. Missing readOnlyHint would help agents understand these tools have no side effects and are safe to call speculatively.
Error handling lacks recovery guidance. No visible try-catch blocks or error messages that tell LLMs what to do next (e.g., 'Publication not found. Available: [list]' or 'ripgrep not installed, install via brew install ripgrep').
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 59 | <=2025-11-25 | v2 |
Full-text search (ripgrep) over the docs. Fast, deterministic, regex-capable.
Semantic search restricted to code examples in the bundle. Best for 'show me an example' queries.
Semantic search across ServiceNow documentation. Best for conceptual or 'how do I' queries. Returns top-k chunks with file paths, headings, and similarity scores.
Parameter descriptions lack format/constraint details. E.g., 'max_chars' has a default (50000) but no guidance on typical values or why an LLM might adjust it. 'regex' is a boolean but no examples of valid regex patterns.
search_docs and search_code accept optional 'publication' and 'product_area' filters but descriptions do not explain what happens if both are provided or if they conflict. Undocumented parameter relationships.