// SCAN REPORT

Smallest

70 tools. 19 can modify or destroy data without limits.

4 destructive tools with no built-in limits. Policy required.

Last updated: 6 May 2026

19 can modify or destroy data

51 read-only

70 tools total

// TOOL MAP

Read (51) Write / Execute (15) Destructive / Financial (4)

// WHAT CAN GO WRONG

Destructive tools (delete_agent, delete_audience, delete_audience_members) permanently delete resources. There is no undo. An agent calling these in a retry loop causes irreversible damage.

Write operations (activate_version, add_audience_members, create_agent) modify state. Without rate limits, an agent can make hundreds of changes in seconds — faster than any human can review or revert.

Execute tools (start_campaign) trigger processes with side effects. Builds, notifications, workflows — all fired without throttling.

// RECOMMENDED RULES

Deny destructive operations

delete_agent:
  rules:
    - action: deny

Destructive tools should never be available to autonomous agents without human approval.

Rate limit write operations

activate_version:
  rules:
    - rate_limit: 30/hour

Prevents bulk unintended modifications from agents caught in loops.

Cap read operations

compare_version_metrics:
  rules:
    - rate_limit: 60/minute

Controls API costs and prevents retry loops from exhausting upstream rate limits.

// ALL 70 SMALLEST TOOLS

DESTRUCTIVE 4 tools

Destructive delete_agent Destructive delete_audience Destructive delete_audience_members Destructive delete_campaign

EXECUTE 1 tools

Execute start_campaign

WRITE 14 tools

Write activate_version Write add_audience_members Write create_agent Write create_campaign Write export_campaign_logs Write invite_member Write publish_draft Write redeem_coupon Write rename_draft Write text_to_speech Write update_agent_config Write update_agent_prompt Write update_billing_alerts Write update_version

READ 51 tools

Read compare_version_metrics Read debug_call Read diff_versions Read duplicate_agent Read get_agent Read get_agent_performance Read get_agent_prompt Read get_agents Read get_attempt_cohorts Read get_audience Read get_audience_members Read get_audiences Read get_auto_reload Read get_billing_alerts Read get_call_counts_by_day Read get_call_outcomes Read get_call_start_distribution Read get_call_volume Read get_campaign Read get_campaigns Read get_concurrency Read get_credit_balance Read get_credit_ledger Read get_credit_usage Read get_daily_summary Read get_dashboard Read get_draft Read get_draft_diff Read get_duration_stats Read get_hourly_performance Read get_invoices Read get_payment_methods Read get_phone_number_trends Read get_phone_numbers Read get_pickup_rates Read get_plans Read get_usage_breakdown Read get_usage_stats Read get_version Read get_voices Read get_weekly_trends Read list_calls Read list_drafts Read list_versions Read make_call Read pause_campaign Read search_audience_members Read test_draft Read test_version Read transcribe_audio Read validate_coupon

// FAQ

Can an AI agent delete data through the Smallest MCP server? +

Yes. The Smallest server exposes 4 destructive tools including delete_agent, delete_audience, delete_audience_members. These permanently remove resources with no undo. Intercept blocks destructive tools by default so they never reach the upstream server.

How do I prevent bulk modifications through Smallest? +

The Smallest server has 14 write tools including activate_version, add_audience_members, create_agent. Set rate limits in your policy file -- for example, rate_limit: 10/hour prevents an agent from making more than 10 modifications per hour. Intercept enforces this at the transport layer.

How many tools does the Smallest MCP server expose? +

70 tools across 4 categories: Destructive, Execute, Read, Write. 51 are read-only. 19 can modify, create, or delete data.

How do I add Intercept to my Smallest setup? +

One line change. Instead of running the Smallest server directly, prefix it with Intercept: intercept -c smallest.yaml -- npx -y @@developer-smallestai/smallest-mcp-server. Download a pre-built policy from policylayer.com/policies/smallest and adjust the limits to match your use case.

// RELATED SERVERS

Other MCP servers with similar tools.

Starter policies available for each. Same risk classification, same one-command setup.

Smallest

Other MCP servers with similar tools.

Let agents act without letting them run wild.