MCP server for SecObserve: triage findings, import scan reports and SBOMs, generate VEX.
SecObserve MCP demonstrates solid tool design with 18 well-structured tools covering CRUD, workflows, and domain-specific operations. All tools have descriptions (avg 180 chars) and input schemas with types. Naming follows verb_noun convention consistently (list_, get_, create_, update_, delete_, assess_, approve_, import_, trigger_, run_). Parameter descriptions are present and mostly actionable. Key strengths: discovery tools (secobserve_list_resources, secobserve_describe_resource) guide agent exploration; workflow tools (assess_observation, bulk_assess_observations) include rich validation and audit-trail guidance; error handling via SecObserveError provides recovery hints. Gaps: output schemas not explicitly documented in source; some parameters lack format/constraint details (e.g., risk_acceptance_expiry_date pattern is present but not all date fields documented); no tool annotations (readOnlyHint, destructiveHint) visible in code; response field naming not verified against input parameter naming for chaining.
Import findings via the SecObserve API import endpoint.
Approve or reject a pending assessment on an observation.
Record a human assessment on one observation: change its severity, status, priority or VEX justification. This is how triage is done. It writes an observation log, so the change is attributable and reversible, and it is what later VEX documents are generated from. Never edit an observation's severity or status with secobserve_update -- that bypasses the log and the approval workflow.
Apply one identical assessment to up to 250 observations by id. The comment is stored on every one of them, so write it to be true of the whole set. Get the ids from secobserve_list with response_format="json" and fields=["id"]; a filter that matches more than 250 rows needs several calls.
Call a named action on a SecObserve resource.
Output schemas not documented in source code. While input schemas are present with types and descriptions, return types and field structures are not explicitly declared. This forces LLMs to infer output structure from descriptions alone, risking misinterpretation of response fields.
Tool annotations (readOnlyHint, destructiveHint, idempotentHint) not visible in source. Tools like secobserve_delete (DESTRUCTIVE) and secobserve_assess_observation (WRITE) lack explicit annotations that would help agents understand side effects and retry safety.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 68 | 2026-07-28+ | v2 |
Create a SecObserve resource.
Generate a VEX (Vulnerability Exploitability eXchange) document.
Delete a SecObserve resource by id.
Read the deployed instance's OpenAPI schema for one resource: filters, fields, enums. This is the authoritative answer to "what can I filter on" and "what does the body need", because it comes from /api/oa3/schema/ on the running backend rather than from a hand-written list. Call it before create/update, and before guessing a filter name.
Fetch one SecObserve resource by id.
Import a scan report or SBOM file into SecObserve.
List SecObserve resources with optional filtering, pagination, and field projection.
List every SecObserve resource this server can reach, with its verbs and named actions. Start here. The output is the vocabulary for secobserve_list / get / create / update / delete / call_action. It is served from a static catalogue and makes no API call, so it is free to call first.
Retrieve metrics for a SecObserve product.
Run a periodic task immediately.
Get the status and version of the SecObserve instance.
Trigger a scan on a SecObserve product.
Update a SecObserve resource by id.
Generic 'body' parameter in create/update/call_action lacks field-level schema. Parameter is typed 'object' with description 'Resource fields as key-value pairs' but does not enumerate required fields, constraints, or valid keys. Agents must call secobserve_describe_resource first, adding latency.
Response field naming not verified against input parameter naming for tool chaining. If secobserve_list returns 'observation_id', downstream tools like secobserve_assess_observation must accept 'observation_id', mismatches force extra lookup calls.
Pagination guidance incomplete. secobserve_list accepts page/page_size but description does not state whether page is 0-indexed or 1-indexed, or what happens when page exceeds total. No mention of total_count or next_cursor in response.