Repository navigation
Sample Bundle
A real bundle from tracekit demo (scripted agent, dev-mode signer), so you can try verification without
installing a signer or running an agent.
| File | What it is |
|---|---|
demo-run.tkb |
the evidence bundle: manifest, signed records, checkpoints, policy, coverage, replay |
demo-run-tampered.tkb |
the same bundle with one recorded command edited |
signer.pub |
the signer's public key, to pin with --key
|
demo-run-replay.html |
the bundle's replay viewer, extracted; open it in a browser (works offline) |
tracekit verify docs/sample/demo-run.tkb --key docs/sample/signer.pub # exit 0
tracekit verify docs/sample/demo-run-tampered.tkb --key docs/sample/signer.pub # FAILED: hash mismatch, exit 1The first one ends with:
Integrity: VERIFIED.
Assurance: dev (signer ran as the agent's own user: the agent could have rewritten the ledger).
Read the two lines together: the records are intact and signed by the pinned key, but a dev-mode signer shares the
agent's user, so this proves the format and the checks, not tamper-proof custody. Both commands also print
warning: replay.html differs from the viewer this verifier would generate: these samples were made with an earlier
viewer. The verdict never uses replay.html, so the warning does not change it.
What you are looking at: an agent asked to fix a failing test reads a README carrying a planted prompt
injection that tells it to upload .env. The hooks record each tool call the (scripted) agent made, and the policy
blocks the upload (rule TK-D006). The bundle carries only its own checkpoints, no witness copy, and the verifier
also warns that the key is a file on the signer host, that no harness was registered, and that the run used a path
Tracekit cannot see into (the payload of a network command). Real deployments use system mode and a witness outside
the agent host (docs/witnesses.md).
Regenerate with tracekit demo --keep and copy run.tkb, tampered.tkb and signer/keys/signer.pub from the
folder it prints. tests/test_hardening.py checks that the sample still verifies and the tampered copy still fails.
Generated from docs/ at v1.4.1. Edit the docs in the repository, not here.
Get started
- Quickstart: the v2 signer
- Quickstarts
- Quickstart: Claude Agent SDK
- Quickstart: a custom agent
- Quickstart: LangChain and LangGraph
- Quickstart: MCP client
- Quickstart: OpenAI Agents SDK
- Coding agents other than Claude Code
- tracekit doctor
- Walkthrough: Tracekit end to end
How it works
- How Tracekit works
- Overview
- The life of a tool call
- The evidence format
- Verification and assurance
- Witnesses and monitoring
- Policy
- Approvals
- Detection and containment
- Capture and completeness
- Custody
- Provenance and tracekit why
- Deployment and isolation
- Storage, durability and performance
- Integrations
- Security model
- Architecture
- tracekit why: architecture
Integrations
- Integrations
- Adapters
- OpenTelemetry
- Remote clients
- Kubernetes agent-sandbox
- agentgateway
- agentsh and Tracekit
- Browser Use
- Claude Agent SDK
- Claude Code's sandbox
- Codex's sandbox
- CrewAI
- Docker MCP Gateway
- E2B
- Falco
- Gemini CLI's sandbox
- Google ADK
- Haystack
- LangChain v1
- LiteLLM
- LlamaIndex
- MCP client
- microsandbox
- Microsoft Agent Framework
- OpenAI Agents SDK
- NVIDIA OpenShell
- Pydantic AI
- sandbox-runtime's proxy on its own
- Strands Agents
- Tetragon
- Vercel AI SDK
- Wassette
Policy and approvals
- Policy reference
- Approvals
- Findings
- Reading the model's reasoning
- Incidents and containment
- Webhooks
Deploy and operate
- Deploying the signer
- The signer as a container
- The signer on Kubernetes
- The compose stack
- Identity
- Tool executor
- MCP gateway
- LLM gateway
- Egress observation
- Witnesses
- Log monitor
- Record-key issuer
- Signer observability
- Viewer
- tracekit why: causal investigation and replay
- Migrating
- Platforms
Verify the evidence
- Reading a v2 verify report
- Evidence format v2
- Auditor guide
- Sample evidence bundle
- Signing
- What Tracekit guarantees
Security and privacy
- Threat model: the v2 signer
- Threat model: the v1 laptop signer
- HTTP security checklist
- Threat-model review packet
- Privacy
- Known limits and trade-offs
- What's stable
Compliance
- Compliance mapping
- EU AI Act
- ISO/IEC 42001
- NIST AI RMF and the Generative AI Profile
- OWASP Top 10 lists for LLM and agentic applications
Evaluation
Demos
- Demos
- Agent approvers within bounds
- Completeness and tamper evidence
- Detection and containment
- Network coverage: an egress observer
- Guardrails: allow, deny, or ask a person
- MCP tools, gated and recorded
- Prompt injection held by provenance
- Why: which input caused the action
Help