Claspine docs, API reference and code checks for AI agents: guides, recipes, errors, upgrades.
Scored across 12 tools
Most tools have sharply distinct purposes: three checkers (code, config, upgrade), separate reference surfaces (HTTP contract, Java types, capability matrix), and retrieval tools (search_docs, get_guide, get_recipe). The closest overlaps are check_upgrade vs upgrade_notes and search_docs vs the get_* family, but descriptions clarify that one summarizes checklists and the other returns raw notes, and that search leads to the get_* follow-ups.
Eight tools follow a clean verb_noun pattern (check_code, get_guide, plan_integration, search_docs), but four are bare noun phrases (api_reference, java_api, upgrade_notes, what_runs_where), mixing two conventions. The noun forms are readable and somewhat defensible as resource accessors, but the set is not uniform.
Twelve tools sits comfortably in the well-scoped range for a platform-documentation server. Each tool maps to a recognizable job (three linters, five reference/retrieval surfaces, an integration planner, an error explainer, an upgrade path), with no filler entries.
Coverage is strong: search, guides, recipes, HTTP and Java references, error explanations, version support matrix, upgrade notes plus an upgrade checker, and an ordered integration planner. The main gap is that nothing helps detect or inventory the user's project stack or current Claspine version beyond check_upgrade's file scan, so plan_integration depends on the agent already knowing the stack.
The published HTTP contract shared by Claspine Atlas and the self-hosted engine (paths under /api/v1). Without path: every operation. With path (and method): parameters, request body and responses with schemas resolved.
Inputs: path, method.
Runs Claspine's Agent Shield rules on files or a unified diff: hardcoded or client-exposed secrets, SQL built from strings, unscoped tenant queries, secrets in logs, buffered uploads, imports of platform internals. Returns file, line, rule and the fix. Code is analysed in memory and not stored. A clean result is not proof of safety; Spring Boot builds also run ProductBoundaryTest.
Inputs: diff, files.
Checks application.yml / application.properties or the engine's environment variables: empty values that switch providers on, sandbox sign-in in STAGING/PRODUCTION, IP addresses as hosts, secrets written into settings, and misspelled setting names. Content is not stored.
Inputs: content (required), format.
Finds the Claspine version in project files (pom.xml, build.gradle, package.json, docker-compose.yml) or takes it as given, and lists the checklist items of every upgrade note up to the target.
Inputs: files, target, current.
What a Claspine error code means and how to fix it: HTTP status, the messages it is raised with, where, and the fix for common ones. Every Claspine error response carries its code.
Inputs: code (required).
The full text of one Claspine guide: steps, code, settings, safety notes and troubleshooting.
Inputs: slug (required).
Copy-ready HTTP calls for a common task on Claspine Atlas or the self-hosted engine, in curl, Node.js, Python, Go or Java, checked against Claspine's API on every build. Without task: the list.
Inputs: form, task, language.
The public Java types of a Claspine capability (its api package) and of the foundation, with methods and record fields, from the current platform build. Without arguments: the capabilities.
Inputs: type, capability.
An exact, ordered plan for adding Claspine to this project: what the person must do first, then each step with the files and settings to change (current release pinned) and how to prove it worked. Call it after you know the stack, new or existing, and the person's decisions (Atlas or self-hosted, whose login, which features), and follow it step by step.
Inputs: form (required), stack (required), project (required), features (required), +4 more.
Search Claspine's guides, recipes, capability table and error fixes with a plain question or keywords. Returns the best matches first, with links; follow up with get_guide or get_recipe.
Inputs: query (required), limit.
Every Claspine upgrade note after one version up to another, oldest first, with what changed, the checklist for product code and what to do before deploying. Read them all before changing the Claspine version.
Inputs: from (required), to.
Which Claspine capabilities exist on Claspine Atlas, the self-hosted engine (any language over HTTP) and self-hosted Spring Boot. Check this before planning any Claspine integration: they differ (for example, Claspine Atlas has no sign-in, permissions or push).
Inputs: capability.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Gemmein docs, API reference, error explainers and a live integration check for coding agents.
Search the official docs of Claude Code, Cursor, OpenAI Codex and MCP, refreshed daily.
Versioned documentation registry and semantic search for AI tools and coding assistants.
Add one secure layer between your agents and this server.