ISO 27001SAFE
Model Context Protocol (MCP) server for ISO27001: A stateful Model Context Protocol (MCP) server that gives Claude a complete ISO 27001:2022 Information Security Management System (ISMS). Ask Claude to run gap assessments, manage risks, generate policies, track evidence, and run audits — all backed
Overview
From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.
Turn Claude into an ISO 27001 compliance assistant — controls, risk register, policies, evidence tracking, SoA generation, and full audit workflows in one local encrypted MCP server.
[](https://npmjs.com/package/iso27001-mcp) [](https://npmjs.com/package/iso27001-mcp) [](https://github.com/Sushegaad/MCP-Server-for-ISO27001/actions/workflows/ci.yml) [](LICENSE) [](https://www.iso.org/standard/27001) [](https://socket.dev/npm/package/iso27001-mcp/overview/0.9.83)
[▶ Live Interactive Demo](https://sushegaad.github.io/MCP-Server-for-ISO27001/) · Guided First-Run Checklist → · Roadmap →
▶ Getting Started — watch on YouTube
Why this exists
ISO 27001 compliance work is typically scattered across spreadsheets, Word docs, ticketing systems, and shared drives. Security teams and consultants spend more time chasing evidence and reformatting documents than actually improving security posture.
`iso27001-mcp` solves this by giving Claude a live, stateful ISMS — all 93 ISO 27001:2022 controls seeded and ready, a real risk register, policy and procedure generators, evidence tracking, audit workflows, and a Statement of Applicability, backed by an encrypted local database that never leaves your machine.
The d
b53b10c3032aOBSERVED · 2026-10-08Connect
Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control. Replace the environment placeholders with a token scoped to the least it needs.
claude mcp add iso27001-mcp --env DB_ENCRYPTION_KEY=${DB_ENCRYPTION_KEY} --env HMAC_SECRET=${HMAC_SECRET} --env MCP_API_KEY=${MCP_API_KEY} -- npx -y [email protected]Exposed tools (52)
29 read · 22 write · 1 destructive. Blast radius: 1 tool can delete or overwrite — an agent that can be talked into calling a tool can be talked into calling this one.
| Tool | Risk | Description |
|---|---|---|
archive_gap_assessment | read | Archive a completed or superseded gap assessment with an optional reason. Omit confirmed or pass confirmed=false to preview without writing; pass confirmed=true to commit (archiving is irreversible via tools). |
compare_versions | read | Show the mapping relationship between a 2013 control and its 2022 equivalent(s), or vice versa. |
create_audit | write | Create an internal ISMS audit record with auditor, planned date, and controls/clauses in scope. Omit confirmed or pass confirmed=false to preview the audit record without writing; pass confirmed=true to commit. |
create_corrective_action | write | Create a corrective action request (CAR) linked to an audit finding with owner and due date. Omit confirmed or pass confirmed=false to preview the CAR without writing; pass confirmed=true to commit. |
create_gap_assessment | write | Create a new gap assessment against ISO 27001:2022 or 2013 controls for a defined ISMS scope. |
create_improvement_opportunity | write | Register a proactive improvement opportunity (ISO 27001:2022 Clause 10.1) with source, priority, owner, and optional target date. Not linked to a nonconformity. |
create_management_review | write | Schedule a new management review (ISO 27001:2022 Clause 9.3) with title, date, and reviewers list. |
create_policy | write | Generate a new ISMS policy document from a Mustache template using organisation_name, scope, owner, and effective_date. Omit confirmed or pass confirmed=false to preview what will be created without writing; pass confirmed=true to generate and save the policy. |
create_procedure | write | Generate a new ISMS procedure document from a Mustache template. Optionally links to a parent policy. Falls back to org profile for organisation_name and scope if not supplied. |
create_risk | write | Register a new information security risk with asset, threat, vulnerability, likelihood (1–5), and impact (1–5). |
create_treatment_plan | write | Create a risk treatment plan (mitigate, accept, avoid, or transfer) with owner, due date, and optionally residual risk scores. |
export_gap_report | read | Export a full gap assessment report in markdown, CSV, or JSON format. |
export_procedure | read | Export a procedure as a markdown document (with related controls appended) or as structured JSON. |
export_soa | read | Export the Statement of Applicability in markdown or CSV format. |
generate_audit_report | read | Export an audit report including findings and CAR status in markdown or JSON format. |
generate_remediation_roadmap | read | Generate a prioritised remediation roadmap from a gap assessment, grouped by risk level and theme. |
generate_risk_register | read | Export the full risk register in markdown, CSV, or JSON format with optional level/status filters. |
generate_soa | read | Generate a Statement of Applicability from a gap assessment, pre-populating inclusion/exclusion for all controls. |
get_control_attributes | read | Retrieve the 2022 attribute tags for a control: information_security_properties, cybersecurity_concepts, operational_capabilities, security_domains. |
import_control_statuses | write | Bulk-update control implementation statuses in a gap assessment from a CSV string. Supports dry_run=true for validation preview. Omit confirmed or pass confirmed=false to preview the import without writing; pass confirmed=true to commit. Headers: control_id, status, notes, na_justification. |
link_github_issue | write | Link an evidence record to an existing GitHub issue by number, or create a new issue from a title and body. |
link_jira_ticket | write | Link an evidence record to an existing Jira ticket by key, or create a new ticket from a summary. |
list_api_keys | read | List all API keys with their metadata (label, role, status, expiry). Never returns key hashes. |
list_clause_requirements | read | List ISO 27001:2022 clause requirements (clauses 4–10), optionally filtered by parent clause. |
list_controls | read | List ISO 27001 controls with optional filters: version, theme, control_type, new_in_2022, cybersecurity_concept, and pagination. |
list_evidence | read | List evidence records for a control, optionally filtered by status (current, stale, expired). |
list_evidence_documents | read | List generated evidence documents with optional filters: template_type, generated_by, control_id, and pagination. |
list_gap_assessments | read | List gap assessments filtered by status (active, archived, all). |
list_improvement_opportunities | read | List improvement opportunities with optional filters (status, source, priority, review_id) and a backlog health rating (excellent/good/fair/needs_attention/at_risk). |
list_management_reviews | read | List management reviews with optional status filter and pagination. |
list_policies | read | List policies with optional filters: status, type, owner, overdue_only, and pagination. |
list_procedures | read | List procedures with optional filters: procedure_type, status, policy_id, overdue_only, and pagination. |
list_risk_acceptances | read | List recorded risk acceptance decisions with optional filters (risk_id, decision) and pagination. |
list_risks | read | List risks with optional filters: risk_level, status, owner, and pagination. |
query_audit_log | read | Query the tamper-evident audit log with optional filters: date range, tool, outcome, role, key_hash, actor_type (ai | human | system). |
record_finding | read | Record an audit finding (NC, observation, or OFI) against a clause or control. Omit confirmed or pass confirmed=false to preview the finding without writing; pass confirmed=true to commit. |
record_review_input | read | Record one of the 7 mandatory Clause 9.3.2 input categories for a management review. Upserts on re-submission; advances status to in_progress on first input. |
record_review_output | read | Record a Clause 9.3.3 output decision (improvement_decision or isms_change_decision) for a management review. |
record_risk_acceptance | read | Record a risk owner |
register_evidence | read | Register an evidence artefact linked to a control with type, source URL, collector, and optional expiry date. Omit confirmed or pass confirmed=false to preview the evidence record without writing; pass confirmed=true to commit. |
revoke_api_key | destructive | Revoke an API key by label, preventing all future use and evicting its live SSE sessions. Omit confirmed or pass confirmed=false to preview without writing; pass confirmed=true to commit. |
search_controls | read | Full-text search across control names, descriptions, and guidance using the FTS5 index. |
set_organization_profile | write | Upsert the singleton organization profile (legal name, jurisdiction, ISMS scope, RACI roles). Used by create_policy and create_procedure to auto-inject org defaults. |
set_risk_methodology | write | Upsert the organisation |
update_control_status | write | Set the implementation status of a control within a gap assessment (implemented, partial, not_implemented, na, not_started). Omit confirmed or pass confirmed=false to preview changes without writing; pass confirmed=true to commit. |
update_corrective_action | write | Update a corrective action |
update_improvement_opportunity | write | Advance an improvement opportunity |
update_policy | write | Create a new version of an existing policy with scope/owner changes, reviewed_by, and change_summary. Requires admin role. Omit confirmed or pass confirmed=false to preview the version bump and metadata diff without writing; pass confirmed=true to commit. |
update_procedure | write | Archive the current procedure version and re-render with updated fields, incrementing the version number. Requires admin role. Omit confirmed or pass confirmed=false to preview the version bump and metadata diff without writing; pass confirmed=true to commit. |
update_risk | write | Update mutable fields of an existing risk (asset, threat, vulnerability, likelihood, impact, owner, status, related_controls). Omit confirmed or pass confirmed=false to preview a field-level diff without writing; pass confirmed=true to commit. |
update_soa_entry | write | Update an SoA entry |
update_treatment_status | write | Update the status and evidence reference for an existing risk treatment plan. Omit confirmed or pass confirmed=false to preview changes without writing; pass confirmed=true to commit. |
Trust audit
SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | PASS |
| L1 | Static analysis of the code | PASS |
| L2 | Instruction surface (what it tells the agent) | PASS |
| L3 | Class-specific surface | WARN |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- declared (3 observation(s))
- Network
- declared (4 observation(s))
- Shell
- none-observed
- Dependencies
- not all pinned
- Secrets in source
- none-found
Findings (25)
revoke_api_key
.nojekyll
.nojekyll
const pkg = require("../../package.json") as { version?: string };import { MIGRATIONS } from "../../src/db/migrations/index.js";import { seedAll } from "../../src/seed/seeder.js";import { McpError } from "../../src/types/errors.js";vi.mock("../../src/db/connection.js", () => ({| `CORS_ORIGIN` | `http://192.168.1.50:3000` | The exact origin your Claude Desktop connects from |
proxy_pass http://127.0.0.1:3001;
@vitest/coverage-istanbul, @vitest/coverage-v8, ajv, ajv-formats, vitest, express
**Server egress:** `iso27001-mcp` itself does not transmit ISMS records to the package author, analytics services, or any AI provider. In local stdio mode the server opens no network listener and make
incident_handling: ['Detection & Identification — monitor SIEM alerts; staff report via [email protected]', 'Initial Assessment — classify severity (P1/P2/P3) within 15 min; P1 triggers war-room', 'Con
> **Server egress:** `iso27001-mcp` itself does not transmit ISMS records to the package author, analytics services, or any AI provider. In local stdio mode the server opens no network listener and ma
- Must report any suspected compromise of their credentials immediately to the security team ([email protected])
> Your Node.js was installed system-wide and `npm install -g` needs write access to a root-owned directory. **Do not use `sudo npm install -g`** — it causes other issues. Instead, install Node via [nv
> **Permission error on Linux/macOS?** Do not use `sudo npm install -g`. Instead, install Node.js via [nvm](https://github.com/nvm-sh/nvm) or [Volta](https://volta.sh) — both install Node into your ho
**Finding:** AWS IAM access advisor review identified 4 developer accounts with persistent `AdministratorAccess` policy attached, providing unrestricted access to all AWS services in the production en
- AWS Config conformance pack deployed with rules covering: S3 public access block, IAM password policy, CloudTrail multi-region enabled, Security Groups unrestricted access, KMS key rotation enabled,
- Read-only guest access (no API key required for public resource URIs in hosted mode)
- **Malicious local user** — a user on the same machine who can read environment variables, file system paths, or process memory
- **stdio transport:** The OS process boundary is the trust boundary. No network interfaces are involved. A process-level attacker (running as the same user) can read environment variables. Operators
- Are responsible for not sharing their access credentials with any other person
6. **Post-departure monitoring:** AWS CloudTrail and system logs are monitored for 30 days post-departure for any access attempts using former credentials
**Pinned toolchain:** Node.js version is pinned via `volta.node` in `package.json` (currently `20.11.0`). Install Volta to get the exact same Node version: `curl https://get.volta.sh | bash`.
Gates applied: no_behavioural_pass.
b53b10c3032afull audit observations/trust-audit/mcp-server/sushegaad__iso-27001.json · Report an issue / request a re-scanAudit history
Every audit this server has had. A grade with a past is a grade somebody is still checking.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-10-08 | b53b10c3032a | SAFE | B | 89 | first audit |
Questions
What is the ISO 27001 MCP server?
Model Context Protocol (MCP) server for ISO27001: A stateful Model Context Protocol (MCP) server that gives Claude a complete ISO 27001:2022 Information Security Management System (ISMS). Ask Claude to run gap assessments, manage risks, generate policies, track evidence, and run audits — all backed
What tools does ISO 27001 expose?
52 in total: 29 read-only, 22 that write, and 1 that can delete or overwrite (revoke_api_key). Every one is listed on this page with its risk.
Is ISO 27001 safe to connect to an agent?
The audit found nothing in the source that contradicts what it says it does, and graded it B (89/100). Grade A is held back for packages that have also passed a sandboxed behavioural run, which is why a clean server reads B. Separately from the audit: 1 of its tools can destroy data, so scope the token you give it to what you actually need.
What credentials does ISO 27001 need?
It reads DB_ENCRYPTION_KEY, HMAC_SECRET and MCP_API_KEY from the environment. Give it a token scoped to the least it needs — an agent that can be talked into calling a tool can be talked into calling it with your credentials.
How does ISO 27001 run?
It speaks sse and stdio, so it runs as a local process your client starts. It is published on npm as iso27001-mcp at 0.9.83.
How current is this page?
The grade is for one exact copy of the source (b53b10c3032a), read on 2026-10-08. The repository is watched and re-audited when it changes.