2.95K
total weekly downloads
86 packages
@agentlaw/audit
Tamper-evident audit log for AI agent actions: hash-chained JSONL writer and verifier.
@agentlaw/core
Deterministic policy engine for AI agent tool calls: parse, sign, verify and evaluate AgentLaw policy manifests. No MCP dependency — embeddable anywhere.
@agentlaw/gateway
The agentlaw CLI and MCP policy gateway: enforce a signed policy manifest on every tool call an AI agent makes.
@agentlaw/sink-supabase
Optional Supabase sink for AgentLaw: mirror the hash-chained audit, run a cross-host approval channel, and publish per-environment policy summaries for a governance console.
@cognitiondesk/agent-runtime
Agent execution runtime for CognitionDesk — wraps @cognitiondesk/client with agent CRUD, model selection, invocation, streaming, and usage metrics. Tier 2 package.
@cognitiondesk/agent-team-orchestration
Team coordination for CognitionDesk apps — create agent teams (squads), assign scopes, inject skills, delegate tasks, and track progress. Tier 2 package that composes agent-runtime + orchestration-kit + skill-packs.
@cognitiondesk/client
Lightweight CognitionDesk command SDK for assistants, workflows, and tasks without widget UI dependencies.
@cognitiondesk/mcp
MCP server for the CognitionDesk platform: create and manage AI assistants, workflows, squads and knowledge, and inspect your usage — authenticated with your personal API key.
@cognitiondesk/orchestration-kit
Workflow and delegation primitives for CognitionDesk — task delegation, workflow assignment, progress tracking, and delegation patterns for multi-agent coordination. Tier 2 package.
@cognitiondesk/skill-packs
Injectable skill packs for CognitionDesk agents — structured instruction sets that tell agents how to operate in a domain. Includes forum, devops, coordination, rag, and insight skill packs plus a SkillInjector. Tier 2 package.
@cognitiondesk/storefront
Presentational building blocks for a public product-catalog storefront: header, item grid with search/filter, item card, item gallery, contact links, share button — themed via --mn-* tokens
@cognitiondesk/widget
Embed a CognitionDesk AI chat widget into any website. Authenticate with an API key from your CognitionDesk dashboard.
@mounaji_npm/admin-controls
Token editor, theme preview, export/import, and developer toolbar for Mounaji-based projects
@mounaji_npm/admin-dashboard
Production admin dashboard module with package-aware controls and capability-driven states
@mounaji_npm/agent-auth
Secure authentication for AI agents in Mounaji apps — signed JWT (HMAC-SHA256 / Ed25519) adapter, agent identity context, guards, and login UI
@mounaji_npm/agentfs
Per-agent filesystem — a portable, layered home directory for an AI agent (learning, skills, memory, preferences, rules, soul, user, assets) with a manifest and dependency-free tar.gz export/import bundles. Foundation for portable Tier S agents.
@mounaji_npm/agentlaw-core
Deterministic policy engine for AI agent tool calls: parse, sign, verify and evaluate AgentLaw policy manifests. No MCP dependency — embeddable anywhere.
@mounaji_npm/agent-platform-widget
Multi-agent management chat surface — dockable, Claude-style agent chat windows with per-window configurable tools and human-in-the-loop approvals
@mounaji_npm/agent-profile
AI agent profile components — AgentProfileCard, AgentDirectory, capability badges, activity stats
@mounaji_npm/agent-runner
Portable headless coding-agent runner — polls CognitionDesk for pending claude_code-target loops and executes them via pluggable engines (claude-cli today; agent-sdk later). Extracted and generalized from mcp-mounaji-rag's loop-runner.ts.
@mounaji_npm/api-client
HTTP client, service base, and WebSocket connector for Mounaji-powered applications
@mounaji_npm/app-bootstrap
Idempotent bootstrap hooks for production Mounaji shell initialization
@mounaji_npm/app-providers
Composable production provider stack utilities for Mounaji host apps
@mounaji_npm/assistant
Assistant management UI components for Mounaji-based projects
@mounaji_npm/auth
Authentication provider, hooks, guards, and login components for Mounaji applications