Full list of Keprix features

Agent runtime

FastAPI core that runs agent turns, model routing, tools, streaming, sessions, and approvals.

Used for: The shared brain behind every Keprix interface.

Command Center TUI

Keyboard-first Textual terminal with live sessions, slash commands, tool cards, review mode, and diagnostics.

Used for: Operators who want full control without leaving the terminal.

Web workspace

Browser shell for chat, documents, CRM, automations, settings, and admin pages.

Used for: Day-to-day operator work in a shared UI contract.

CLI

keprix CLI for start, setup, memory, security audit, and operator tasks.

Used for: Scripted ops, install, and headless administration.

OpenAI-compatible API

REST and OpenAI-style chat completions against your self-hosted runtime.

Used for: Connecting external clients and apps to Keprix.

Chat

Streaming agent sessions with models, tools, slash commands, and voice input.

Used for: Primary conversational work with the agent.

Documents

Upload, index, share, and export workspace files.

Used for: Keeping source material next to the agent.

Notes

Linked notes and references inside the workspace.

Used for: Capturing decisions and working context.

Tasks

To-do, in progress, and done tracking for operator work.

Used for: Turning agent outputs into tracked follow-ups.

Calendar

Month, week, and day views with schedule-aware agent context.

Used for: Planning work and booking-related flows.

Email

IMAP inbox, AI triage, and send APIs behind Channel Shield where enabled.

Used for: Reading and acting on mail safely.

Contacts

People records, sync, and communication preferences.

Used for: CRM and outreach contact identity.

Gallery

Generated and uploaded images in one browser.

Used for: Reviewing media produced by agents and tools.

Vical booking

Booking pages, reschedule/cancel flows, and Soft Wall enroll hooks.

Used for: Public scheduling tied to CRM and outreach.

Notifications

In-app inbox, digests, and optional external SMTP alerts.

Used for: Keeping operators aware of approvals and system events.

Agent OS

Action boards, run ledgers, client kits, promote flows, and self-improvement loops.

Used for: Operating agents as a managed workflow system.

Mutation engine

Proposes new tools or code, sandboxes and tests them, then waits for human approval.

Used for: Closing capability gaps without silent live installs.

Self-coding agent

Long-horizon coding sessions with diffs, tests, and risk review.

Used for: Repo changes that stay reviewable.

Playbooks

Visual and YAML workflows with runs, approvals, schedules, and triggers.

Used for: Repeatable automations with human checkpoints.

Cron jobs

Scheduled agent tasks with run history.

Used for: Recurring maintenance and outbound jobs.

Agent Studio

Build and publish manifest-driven agent experiences.

Used for: Packaging reusable agent apps.

Agent Apps

Install, run, and schedule packaged agent apps.

Used for: Productized agent workflows for a workspace.

Skills and plugins

Skills Hub, skill-first execution, and extension plugins.

Used for: Adding domain behavior without forking core.

Improvement loop

Proposes skill, tool, and workflow upgrades through review.

Used for: Continuous self-improvement under Soft Wall / approvals.

Agent teams

Multi-agent collaboration patterns on one runtime.

Used for: Delegating work across specialized agents.

Control Center

Operator cockpit for runtime status and guided actions.

Used for: Seeing what the fleet of agents is doing.

Long-term memory

Structured, workspace-scoped memory with semantic search.

Used for: Agents that remember facts across sessions.

Brain graph

Graph views of entities, memories, and relationships.

Used for: Exploring how knowledge connects.

RAG pipelines

Index and retrieve documents, workspace data, and codebase self-knowledge.

Used for: Grounded answers from your corpus.

Self-knowledge

Indexed product docs so Keprix can answer what it can do.

Used for: Operator and agent questions about the platform.

Graphiti bridge

Optional knowledge-graph bridge for richer entity memory.

Used for: Advanced memory topologies when enabled.

Agentic CRM

Leads, contacts, accounts, deals, lists, ICP, SLA, inbox, workflows, analytics, and settings in one CRM surface.

Used for: Running go-to-market work with agent assistance.

Discovery and enrich

Find and enrich prospects from directories, sheets, and licensed sources.

Used for: Building qualified pipelines without leaving Keprix.

Companies House

UK company lookup and enrichment against the official API.

Used for: B2B research and compliance-aware prospecting.

Decision-maker identification

Fills in a missing contact name from a lead's company officers, ranked by role, with a web-search fallback. Fill-empty-only - never invents a name or overwrites one you already have.

Used for: Turning a company-only lead into a named contact before outreach.

OSINT public-footprint enrichment

Optional, policy-gated lookup of a contact's public online footprint. Off by default, requires explicit opt-in plus a lawful-use acknowledgement, and never surfaces breach data or recovery emails/phones.

Used for: Extra context on a lead when you've explicitly turned it on for that workspace.

Outreach

Campaigns, sequences, outbox, replies, suppressions, and deliverability.

Used for: Controlled multi-channel outbound.

Soft Wall

Approve-then-retry gate for sensitive CRM and mutation actions before they execute.

Used for: Keeping high-risk writes human-reviewed.

CRM compliance

Contactability, suppressions, merges, tracking privacy, and data quality tools.

Used for: Safer outreach and cleaner CRM data.

Health and social care pack

Domain pack patterns for regulated health/social outreach contexts.

Used for: Sector-specific CRM and messaging constraints.

Channel Shield

Scan, policy-check, sandbox, quarantine, and summarize inbound email and messaging.

Used for: Stopping unsafe content before it enters workflows.

Messaging channels

Telegram, Discord, Slack, WhatsApp (via providers), webhooks, and personas.

Used for: Talking to the agent where your team already works.

Voice

Voice input, templates, wake words, and receptionist-style flows.

Used for: Hands-free and phone-adjacent agent use.

Vault and credential proxy

Encrypted secrets with injection that keeps credentials out of prompts.

Used for: Safe tool access to third-party APIs.

Review gateway

External human review tokens for sensitive actions.

Used for: Cross-team approval of high-risk work.

Tool and resource ACLs

Scope which tools and resources each actor can use.

Used for: Least-privilege agent and operator access.

Governance and evidence packs

Governance providers, pack gates, and evidence for audits.

Used for: Regulated or enterprise review trails.

Multi-tenancy and isolation

Workspace and tenant boundaries for memory, CRM, and credentials.

Used for: Keeping customer data separated.

Scout / governance connector

Optional sanitized signals and control actions to a governance provider.

Used for: External monitoring without giving up self-hosting.

Deep research

Cited research projects with datasets and reproducibility hooks.

Used for: Long-form investigation with sources.

Compare models

Blind A/B evaluation of LLM providers.

Used for: Choosing models with evidence, not vibes.

Evals

Evaluation cases, traces, and result drawers.

Used for: Regression testing agent behavior.

Analytics workspace

Usage, cost, and operational analytics panels.

Used for: Understanding spend and agent activity.

Browser automation

Harnessed browser sessions for agent web tasks.

Used for: Sites that need real browser interaction.

MCP connectors

Model Context Protocol servers as first-class tool sources.

Used for: Plugging external tools into the agent.

A2A

Agent-to-agent provider patterns.

Used for: Interoperability with other agent runtimes.

Universal Sidecar

Public contract for product sidecars: health, capabilities, pairing, jobs, events, and kill switch.

Used for: Embedding Keprix beside Clinicom, Fleetz, AbbiS, and other products.

Domain packs

Installable knowledge and product packs (hub, clinical gates, product sidecars).

Used for: Sector and product-specific capabilities.

SDK

Python and TypeScript clients for the App Foundation and APIs.

Used for: Building on Keprix from other codebases.

Google Workspace

OAuth-backed Workspace integrations where configured.

Used for: Mail, calendar, and drive-adjacent workflows.

n8n sidecar

Optional n8n alongside Keprix via MCP bridge.

Used for: Teams that already automate in n8n.

Admin dashboard

Users, tools, channels, mutations, fleet, and readiness.

Used for: Instance administration.

Feature flags and modules

Progressive UX flags and module inventory.

Used for: Rolling out surfaces without breaking the nav.

Quotas and usage

Token and usage budgets with operator visibility.

Used for: Controlling cost per actor or workspace.

Billing hooks

Optional Stripe wiring for commercial surfaces; Community stays free.

Used for: Managed or donation flows without gating OSS.

Hot backup

Snapshot, verify, and restore archives.

Used for: Disaster recovery for self-hosted instances.

Localization

Corrections and metrics for locale-aware copy.

Used for: Tuning language for your operators.