{"id":"feenlace-mcp-1c","name":"mcp-1c","af_score":46.8,"security_score":38.8,"reliability_score":13.8,"what_it_does":"MCP server (Go binary) that integrates AI assistants with 1C:Enterprise by exposing MCP tools to inspect 1C configuration/metadata and perform limited query and code-related operations over a local 1C HTTP service (and an installed 1C extension).","best_when":"You have a 1C environment with an HTTP service you control, and you want an MCP-capable AI client to generate/analyze BSL code using your real configuration metadata, ideally with local models or within your network.","avoid_when":"You cannot control or secure the 1C HTTP endpoint, you need strong guarantees about auth, logging/PII handling, or you must rely on undocumented/undemonstrated HTTP security behaviors for a public-facing deployment.","last_evaluated":"2026-03-30T15:18:28.218972+00:00","has_mcp":true,"has_api":false,"auth_methods":["MCP client-to-server transport via local process/command args (no auth described for MCP itself)","1C HTTP service credentials via --user/--password for MCP-1C to access 1C HTTP service (if using those flags)"],"has_free_tier":true,"known_gotchas":["Tool behavior may depend on correct setup of the 1C HTTP service and installation of the 1C extension via --install; failures could look like missing metadata or connectivity issues.","For execute_query/validate_query, the docs state SELECT-only—agents may need to constrain themselves to read-only queries.","search_code has multiple modes (smart/regex/exact); agents should choose the appropriate mode to avoid unexpectedly expensive or overly broad queries.","Advanced beta and paid tools are not part of the open MCP set; agents should detect which tools are available in the running server/version."],"error_quality":0.0}