Fabraix Red Team Playground
Free · Open source
AI security researchers
Enclawed
FreemiumSelf-Hosted
Your agent moves money, touches production databases, or reads patient records — and when something goes wrong, you have no tamper-evident record of what it decided or why, and no mechanism that could have stopped it mid-action. Enclawed is a security and compliance wrapper that gates every step of an agent's decision loop and anchors every action to a hash-chained audit trail that auditors can actually examine.
Similar tools
Fabraix Red Team PlaygroundFreeFabraix Red Team Playground is an open-source, MIT-licensed testing environment structured…
EclusiaFreemiumEclusia runs locally on your machine and replaces names, IBANs, phone numbers, API keys, and…
Patronus ScannerFreemiumPatronus AI positions itself as evaluation and monitoring infrastructure for production LLM…
ArgotFreeArgot runs statistical checks against your repository's own commit history to flag code that…
Summary
Enclawed sits between the four phases every agent runs — read, think, act, report — and installs enforcement gates at each transition: a prompt shield on input, plugin admission control before tools fire, egress filtering and data-loss prevention before output leaves, and a multi-witness audit log that cannot be edited after the fact. A boot-time integrity check confirms the runtime has not been tampered with before any action executes. The vendor states coverage maps to SOC 2, HIPAA, FedRAMP Moderate, FIPS 140-3, and several CMMC and NIST frameworks out of the box, with extension to custom audit requirements. The hardware root-of-trust device — a physical anchor that holds keys and must be present for agents to run — is available for preorder with FCC certification pending. Teams that need a purely software-layer solution can start with the open-source MIT-licensed core; the hardware and full compliance evidence tiers are paid-only features.
Bottom line: Pick Enclawed when you are deploying an agent on regulated data and your auditor needs machine-readable proof of every action it took — plan a different approach if your agent workload runs in an air-gapped environment where a hardware preorder shipping on FCC certification timing is an unacceptable dependency.
Find your fit
Before you choose Enclawed
More options in Guardrails & Safety. Compare pricing, platforms, and features side by side.
Eclusia
Freemium
Legal professionals handling confidential cases
Pros
- Gates tool calls before execution — not just text output — so an agent cannot invoke an unauthorized plugin or write to a database even if the language layer passes inspection.
- Hash-chained, multi-witness audit log that cannot be edited after the fact, which means your auditor gets machine-readable evidence rather than log files you assembled the night before the review.
- Boot-time integrity verification confirms the runtime has not been tampered with before any agent action executes, so a compromised environment fails loudly at startup instead of silently mid-task.
- Self-hosted deployment option, so client data and agent decision records stay inside your own infrastructure rather than transiting a third-party cloud.
- Prompt shield covers binary-channel injection vectors — base64 payloads, steganographic images, audio-encoded instructions — that text-only input rails pass through untouched.
Cons
- The hardware root-of-trust device is in preorder status with FCC certification listed as expected by a specific date in 2026; any team whose security architecture requires the physical key anchor cannot ship to production on this component until certification clears, and that timeline is outside the buyer's control.
- The open-source MIT-licensed tier gives you the control gates but not the packaged compliance evidence output; teams using the free core to satisfy a formal audit will spend engineering time mapping OSS telemetry to framework control requirements themselves — at which point the 'compliance ready' value proposition requires the paid tier to actually deliver.
- No API is available, which means integrating Enclawed's controls or audit data into an existing security information pipeline — SIEM ingestion, ticketing triggers, automated compliance dashboards — requires direct file or runtime integration rather than a webhook or REST call; teams with mature SecOps tooling will hit this wall before they finish the proof of concept.
About
- API Available
- No
- Self-Hosted
- Yes
- Last Updated
- September 25, 2026
Best For
Who it's for
- Regulated enterprises (banks, hospitals, federal agencies)
- Teams deploying agents on production systems or client data
- Auditors requiring machine-readable compliance evidence
- Users needing FIPS-aligned cryptographic boundaries
What it does well
- Securing agents that move money or access sensitive data
- Generating tamper-evident audit trails for regulatory audits
- Enforcing mandatory access control and DLP on agent actions
- Boot-time integrity verification for agent runtimes
Frequently Asked Questions
- Is Enclawed free?
- Enclawed offers a free option alongside paid upgrades. Check the current allowances and terms before choosing a plan.
- Can I self-host Enclawed?
- Yes. Enclawed supports self-hosting on your own infrastructure.
Curated lists that include this category
When an agent touches money or records, you lack any stop or proof
An agent that moves funds, writes to production databases, or reads patient records can act without a record of why it chose that step or a way to block the move. Enclawed places gates at each phase of the read-think-act-report loop.
Enforcement points
A prompt shield checks input. Plugin admission control reviews tool calls before they run. Egress filtering and data-loss prevention scan output. A multi-witness audit log records every step in a hash-chained trail that cannot be changed later. A boot-time integrity check verifies the runtime before any action starts.
Compliance scope
The vendor states the controls map to SOC 2, HIPAA, FedRAMP Moderate, FIPS 140-3, and several CMMC and NIST frameworks, with extension available for custom requirements. The self-hosted option exists.
Use cases
Teams secure agents that move money or access sensitive data, generate tamper-evident audit trails for audits, enforce access control and DLP on actions, and run boot-time integrity checks on runtimes.
Trade-offs
The hardware root-of-trust device remains in preorder with FCC certification listed as expected by a specific date in 2026, so teams that need the physical anchor cannot ship until that date passes. The MIT-licensed open-source tier supplies the control gates but omits packaged compliance evidence output, leaving mapping work to the user.
Who it is for / who should skip it
Best for regulated enterprises such as banks, hospitals, and federal agencies, teams running agents on production systems or client data, auditors needing machine-readable evidence, and users seeking FIPS-aligned boundaries. Skip it if you need the hardware anchor before the listed certification date or want ready-made audit packages from the free tier without extra engineering effort. The gates stop tool calls before execution, so an unauthorized plugin cannot fire even if the language layer approves it.
Contribute
- Work on Enclawed? Claim this listing.
- Embed a badge for your README or site.
Sign in to rate this tool or suggest an edit.