Tool overview
Armature is listed under Data Analytics AI tools.
What is Armature?
Armature captures sessions that users run through Claude, ChatGPT, Cursor, Codex, Gemini CLI, and other clients against an MCP server or agent-connected product. It reconstructs intent, reasoning, tool calls, success, failures, and unsupported use cases for product teams.
Best for
Product teams measuring how external AI agents experience their software
Who is it for?
Decision note
Confirm session volume, retention, PII controls, supported backend languages, overage billing, security, SSO, audit logs, custom support, and production data handling.
Key features
Agent-session capture through MCP backends
User-intent and reasoning reconstruction
Automatic success scoring and issue grouping
Full session replay with tool calls
Use-case volume and success analytics
Default PII and secret redaction
Use cases
Understanding how agents use a product
Finding loops and unsupported workflows
Replaying failed agent sessions
Prioritizing agent-experience fixes
Pros
- Free plan includes 1,000 sessions monthly
- Supports any client reaching the MCP server
- Redacts PII and secrets before storage
Limitations
Armature observes only instrumented sessions and can misclassify intent, success, or root cause. Teams must verify redaction, retention, sampled coverage, user consent, and every product decision derived from analytics.
Pricing details
Billing options
Pricing note
The Free plan includes 1,000 sessions per month, unlimited projects and users, seven-day retention, and core analytics. Usage beyond the included amount costs $50 per 1,000 sessions. Custom plans add SSO, audit logs, retention, onboarding, support, and SLA terms.
Supported languages
- TypeScript
- Python
- Go
Integrations
MCP servers
Claude
ChatGPT
Cursor
Codex
Gemini CLI
Please log in to join the discussion.