Agent Decision Infrastructure

Prove what your AI agents decided and why.

Set a password, deploy to Vercel, paste one env var into your agent. No OAuth app required. Your first governed decision shows up in the dashboard in under 10 minutes.

177+ SDK methods 29 categories

Open-source. Self-hosted. Built for teams shipping AI agents to production.

Policy guardDecision audit trailRisk signalsCompliance mappingAssumption trackingHuman-in-the-loop
Deploy Free Live Demo

A dashboard you will actually use

One screen for actions, risk, approvals, messages, and fleet context. Click the screenshot to view fullscreen.

RealtimeHITL approvalsGuard policiesSwarm map
DashClaw Mission Control - strategic overview of your agent fleet
View full gallery →

Secure your first agent in 5 minutes

Four steps from install to full decision governance.

1

Instrument in 60 seconds

npm install dashclaw
# or
pip install dashclaw

Zero dependencies. Works with Node.js and Python agents.

2

Guard before you act

const decision = await dc.guard({
  actionType: 'deploy',
  riskScore: 85,
});

Intervene with real-time policy checks.

3

Score with your own quality bar

// Define what "good" means for your use case
await dc.createScoringProfile({
  name: 'deploy-quality',
  action_type: 'deploy',
  dimensions: [
    { name: 'Speed', weight: 0.3, data_source: 'duration_ms',
      scale: [
        { label: 'excellent', operator: 'lt', value: 30000, score: 100 },
        { label: 'poor', operator: 'gte', value: 120000, score: 20 },
      ]},
    { name: 'Reliability', weight: 0.4, data_source: 'confidence',
      scale: [
        { label: 'excellent', operator: 'gte', value: 0.9, score: 100 },
        { label: 'poor', operator: 'lt', value: 0.7, score: 25 },
      ]},
  ],
});

// Or let DashClaw suggest thresholds from your real data
const suggestions = await dc.autoCalibrate({ lookback_days: 30 });

Custom weighted quality scoring with auto-calibration.

4

Record and Learn

with claw.track(action='deploy'):
  # ... decisions stream to
  # dashboard in real-time

Decisions, policy checks, and signals stream in real-time.

Everything you need to trust your agents

Built for teams running autonomous AI agents in production.

Prove Every Decision Your Agents Make

Every action recorded with reasoning, assumptions, and policy compliance. A live decision ledger streams everything so you can prove why any agent acted.

Enforce Policies Before Agents Act

Semantic guard policies intercept intent before execution. Natural language rules evaluated in real time. No hard-coded checks to maintain.

Attribute Every Dollar to a Decision

Cost-per-decision and burn rate by goal in real time. Financial accountability for every autonomous operation.

Human-in-the-Loop Decision Gates

Approval workflows pause risky decisions for human review. Agents request permission, not forgiveness.

Built-in Evaluation Framework

5 built-in scorer types (regex, keywords, numeric range, custom, optional LLM judge) to track output quality automatically.

Learning Velocity & Maturity

Track how fast your agents are improving. 6-level maturity model from Novice to Master based on real performance data.

Drop-In SDKs

Connect any agent in minutes. Zero-dependency Node.js and Python clients with native adapters for OpenClaw, CrewAI, AutoGen, and LangChain.

Prompt Registry

Version-controlled prompt templates with mustache variables and instant rollback. Stop hardcoding prompts in your agent code.

User Feedback Loop

Collect structured ratings with auto-sentiment detection and 6-category auto-tagging. Close the loop between users and agents.

Behavioral Drift Detection

Statistical baselines and z-score alerts catch when agent behavior deviates from the norm. Detect logic drift early.

Compliance Export Bundles

One-click exports for SOC 2, NIST AI RMF, EU AI Act, and ISO 42001. All your governance evidence, packaged.

Scoring Profiles

User-defined weighted quality scoring with auto-calibration from real data. Risk templates replace hardcoded agent risk numbers with transparent, editable rules.

Verified Agent Identity

Know which agent took which action. RSA signature verification ensures accountability at every step.

Keep Agent Memory Clean

Detect stale facts, repetition loops, and context bloat before they cause bad decisions.

What makes DashClaw different?

Governance that doesn't slow you down.

No-Code Policy Engine

Define guardrails in natural language and enforce them across all agents instantly. No hardcoded checks or messy conditional logic.

Learning Velocity

DashClaw doesn't just log actions; it tracks how fast your agents are improving and where they're still struggling.

Financial Attribution

Every dollar spent on tokens is attributed to a specific decision and goal. Know exactly what your AI budget is buying.

Operator-Defined Quality

You define what "good" means. Weighted multi-dimensional scoring with auto-calibration from your real data. Risk templates replace guesswork with transparent rules.

Complete platform scope

DashClaw is more than a dashboard. It is a full platform spanning control plane UX, APIs, data contracts, realtime transport, SDKs, and CI governance.

Control Plane + Dashboard

Onboarding, team roles, approval queue, risk signals, live action views, and platform health cards.

API + Data Layer

Typed repository boundaries, route contract governance, maturity labels, and OpenAPI drift checks.

Realtime Runtime

Broker-backed SSE fanout, reconnect with Last-Event-ID replay, and cutover health controls.

SDK + Tooling

Node and Python SDKs, CLI toolkit, parity test suites, and docs/CI governance.

Production hardening shipped

Agents That Learn From Their Mistakes

Every completed action is scored and turned into recommendations. Your agents get better without manual retraining.

Explore

Data Layer You Can Trust

SQL drift checks and contract tests run in CI. No silent regressions reach production.

Explore

APIs That Never Break Silently

OpenAPI drift checks catch contract changes before they ship. Your integrations stay stable.

Explore

SDKs That Stay in Sync

Node and Python SDKs are tested against the same contract fixtures. Feature parity is enforced, not hoped for.

Explore

Always-Fresh Recommendations

Automated background jobs keep learning data current. No manual cron jobs to manage.

Explore

Compliance Without the Spreadsheets

Map your guardrails to SOC 2, ISO 27001, GDPR, and NIST AI RMF automatically. Generate audit-ready reports in one click.

Explore

The Right Agent for Every Task

Tasks automatically route to the best-fit agent based on skills, load, and track record. Failed tasks retry and escalate.

Explore
AI Governance Suite

The only agent platform with built-in compliance

Most platforms stop at logging. DashClaw ships decision enforcement, regulatory compliance mapping, and intelligent task routing. All auditable, all testable, all live in the demo.

Compliance Engine

  • SOC 2, ISO 27001, GDPR, NIST AI RMF, EU AI Act
  • Control-level gap analysis with remediation priorities
  • Audit-ready reports in Markdown or JSON
  • Live enforcement evidence from guard decisions
Explore Compliance

Policy Testing & Proof

  • Run tests against all active guard policies
  • Per-policy pass/fail breakdown with diagnostics
  • Generate compliance proof reports on demand
  • Import pre-built policy packs (enterprise, SMB, startup)
Explore Policy Testing

Intelligent Task Routing

  • Skill-based agent matching with scoring
  • Real-time load balancing and health verification
  • Urgency-aware queue with retry and escalation
  • Full agent registry with capability tracking
Explore Task Routing

Every feature works in the demo. No signup required.

Launch Demo
177+ methods across 29 categories

One SDK. Full decision governance.

Install from npm or pip. Zero dependencies. Native adapters for OpenClaw, CrewAI, AutoGen, and LangChain. Decision recording, policy enforcement, assumption tracking, handoffs, messaging, and more.

npm packageNode.jsESM + CJSZero Dependencies
View full SDK docs
// govern your agent
import { DashClaw } from 'dashclaw'
const claw = new DashClaw({
apiKey: process.env.DASHCLAW_API_KEY,
agentId: 'my-agent',
})
// check guard before acting
const { decision } = await claw.guard({
action_type: 'deploy',
risk_score: 85,
})
// record a governed decision
await claw.createAction({
action_type: 'deploy',
declared_goal: 'Ship auth service',
})
// create a session handoff
await claw.createHandoff({
summary: 'Completed auth system',
key_decisions: ['JWT over sessions'],
})
// bulk sync agent state
await claw.syncState({ goals, learning, snippets })

7 Decision Integrity Signals

Automatic detection of autonomy breaches and logic drift. Zero configuration.

Autonomy Spike

SIGNAL-01

Agent taking too many actions without human checkpoints

High Impact, Low Oversight

SIGNAL-02

Critical actions without sufficient review

Repeated Failures

SIGNAL-03

Same action type failing multiple times

Stale Loop

SIGNAL-04

Open loops unresolved past their expected timeline

Assumption Drift

SIGNAL-05

Assumptions becoming stale or contradicted by outcomes

Stale Assumption

SIGNAL-06

Assumptions not validated within expected timeframe

Stale Running Action

SIGNAL-07

Actions stuck in running state for over 4 hours

Production-ready operations

Team management, audit trails, webhooks, and more. Built in from day one.

Team Management

Invite your team in seconds. Role-based access keeps operators in control and agents accountable.

Decision Risk Notifications

HMAC-signed webhooks and email alerts fire when decision integrity signals breach thresholds. No more checking dashboards.

Full Audit Trail

Every action is logged with actor, timestamp, and reasoning: ready for compliance audits and debugging.

Ship in 10 Minutes

Four steps: create workspace, generate key, install SDK, send first action. That's it.

Built for Multi-Tenant

Full org isolation out of the box. Each team gets their own agents, keys, and settings.

30+ CLI Tools

Run agent ops locally with Python CLI tools. Push results to the dashboard when you're ready.

30+ Python CLI tools

Local Agent Toolkit

Python CLI tools that run alongside your agent. Local-first with SQLite storage. Add --push to sync anything to your dashboard.

Learning & Decisions

Log decisions, lessons, and outcomes. Track what worked and why.

learner.py log "Used JWT" --push

Context & Handoffs

Key points, threads, and session continuity documents.

context.py capture "Dark theme" --push

Memory & Health

Scan memory files, track entities, detect stale facts.

scanner.py scan ~/.agent/memory --push

Goals & Relationships

Goal milestones, contacts, interactions, and follow-ups.

goals.py add "Ship auth" --push

Security & Audit

Outbound content filtering, session isolation, audit logging.

outbound_filter.py scan message.txt --push

Automation & Snippets

Reusable code snippets with search, tags, and use tracking.

snippets.py add "retry logic" --push
View full toolkit docs

Govern your first agent in 5 minutes

No OAuth setup required to get started. Set one environment variable, deploy, and connect your first agent. Open-source and self-hosted.

Deploy Free Live Demo