Crucible
Crucible is a self-hosted security tool for stress-testing AI agents against attacks such as prompt injection, goal hijacking, and tool misuse before product…
What Crucible is
Crucible is a self-hosted security tool for stress-testing AI agents against attacks such as prompt injection, goal hijacking, and tool misuse before production. It evaluates tool calls, side effects, and outcomes rather than judging generated text alone. The site lists 11 modules and 153 attack vectors, covering eight of ten OWASP Agentic risks, with two in active development. Crucible says it runs 90 attack payloads in under 60 seconds and returns an A–F grade with severity, OWASP mapping, and reproduction steps. It supports LangChain, CrewAI, AutoGen, raw MCP endpoints, and agents with an HTTP or MCP interface. A GitHub Actions integration can fail a build when the security score falls below a configured threshold. The CLI installs with pip and supports Python 3.10 or later; the site says no Docker, cloud signup, or API key is needed to start. Local scanning requires no signup and has no telemetry by default. The core engine and test modules are Apache 2.0 licensed and free, with no security-test feature gating.
Good
- Covers 153 attack vectors across 11 modules
- Evaluates tool calls, side effects, and outcomes
- Supports LangChain, CrewAI, AutoGen, MCP, and HTTP agents
- Local scanning needs no signup and has no telemetry by default
- Core engine and test modules are Apache 2.0 and free
Watch out
- Two of ten OWASP Agentic risks are still in development
- Requires Python 3.10 or later
Who it suits
Crucible suits individual builders and open-source projects that want local agent security checks, as well as teams shipping agents to production. Its CI integration is relevant to users who want builds to fail when a configured security-score threshold is missed.
Verdict
Crucible provides local, self-hosted checks and graded reports for agent behavior under attack. Its free Community plan includes the CLI and listed security tests; Team and Enterprise add features but have no stated prices.
Plans and prices
From Crucible's own pricing page, checked 5 Oct 2026.
| Plan | Price | What it gives you |
|---|---|---|
| Community | Free | For individual builders and open-source projects · Full CLI & all 8 test modules · 153 attack vectors across 11 modules · Local & CI reports (JSON/HTML) · Community Discord support |
| Enterprise | Priced on request | For orgs with compliance & scale requirements · Everything in Team · SSO, audit logs, on-prem / VPC deploy · Custom attack modules & SLAs · Dedicated security engineering support · Custom |
| Team | Priced on request | For teams shipping agents to production · Everything in Community · Drift monitoring & Slack alerts · Regression tracking · Hosted dashboards · Priority support · Early Access |
The facts
Each one read from Crucible's own pages, with the date it was checked.
What it does
- Attack coverage
- The site lists 11 modules and 153 attack vectors, with 8 of 10 OWASP Agentic risks covered and two in active development.checked 5 Oct 2026
- CI/CD
- The GitHub Actions integration can fail a build when an agent's security score drops below a configured threshold.checked 5 Oct 2026
- Install requirements
- The CLI installs with pip, supports Python 3.10+, and the site says no Docker, cloud signup, or API key is needed to start.checked 5 Oct 2026
- Integrations
- It supports LangChain, CrewAI, AutoGen, raw MCP endpoints, and agents with an HTTP or MCP interface.checked 5 Oct 2026
- Intended users
- The Community plan is for individual builders and open-source projects, and Team is for teams shipping agents to production.checked 5 Oct 2026
- License
- The core engine and test modules are described as Apache 2.0 and free, with no feature gating on security tests.checked 5 Oct 2026
- Privacy
- The site says local scanning requires no signup and has no telemetry by default.checked 5 Oct 2026
- Purpose
- Crucible stress-tests AI agents against attack vectors including prompt injection, goal hijacking, and tool misuse before production.checked 5 Oct 2026
- Reports
- Scans produce an A–F graded report with severity, OWASP mapping, and reproduction steps.checked 5 Oct 2026
- Scan speed
- Crucible says it runs 90 attack payloads against an agent in under 60 seconds.checked 5 Oct 2026
- Support
- The Community plan lists Discord support, while Team lists priority support and Enterprise lists dedicated security engineering support.checked 5 Oct 2026
The company
- Maker
- The team page names Khalid Saifullah Sayyed as Founder & Core and says the team was built in 2026.checked 5 Oct 2026
Where it runs
- Testing approach
- Crucible evaluates actual agent tool calls, side effects, and outcomes rather than only generated text.checked 5 Oct 2026
Alternatives to Crucible
- 1
OpenSecureAI Scanner
Free planOpen sourceWebAPILinuxmacOS8.1From $49 a month - 2

PromptGuard
Free planWebAPIExtensionLinux7.9From $19 a month - 3

Promptfoo
Free planOpen sourceWebAPILinuxmacOS7.8Free plan - 4

Giskard
Free planOpen sourceWebAPILinuxSelf-hosted7.6Free plan - 5

AIRTA Red Team
Free planOpen sourceWebAPILinuxmacOS7.5Free plan - 6
CyberSecEval
Free planOpen sourceWebAPILinuxmacOS7.5Free plan
