Superagent review

Superagent is a security layer for software, AI agents, applications, and infrastructure—not an agent builder. Its product description says it runs security checks around the clock, scans code, tests for weaknesses, and sends findings into existing workflows.

The product covers four main areas: reviewing pull requests, red teaming applications and agents, checking whether dependency vulnerabilities affect your app, and applying guardrails to agent inputs and actions. Context guardrails screen external content before an agent reads, follows, or installs it. Runtime guardrails monitor agents and can block supported risky actions using rules evaluated on each host.

There is also Security-One-27B, which scores security events and returns probabilities that an application can use to route decisions. That makes Superagent relevant to teams adding security checks around development or agent workflows. It does not, from the published product details, replace the tools that build or run those workflows.

The main advantage is breadth: Superagent puts code review, dependency investigation, red teaming, and agent safeguards under one product and API/MCP access. The pricing page lists every product and API/MCP access on all plans, including the free plans, so teams can start with a narrow use case rather than buying separate product tiers.

The checks also map to practical points in a workflow:

• PR scanning checks code changes, dependencies, and contributors before merge.
• Dependency Validation investigates which vulnerabilities affect your app and which updates fix them; public investigations are free.
• Red teaming investigates applications, code, agents, and packages for exploitable weaknesses.
• Context and runtime guardrails cover different risks: what agents encounter and what actions they take.
• Security-One-27B returns probabilities applications can use to route security decisions.

The pricing page lists unlimited seats on every plan. For public repositories, the Open source plan includes unlimited usage at no charge. That can make the product easier to evaluate across a team, although private-repository checks still draw from a monthly credit allowance.

The main constraint is credits. Free includes 100 private credits a month; Starter is $499/month for 500, Growth is $1,499 for 2,000, and Scale is $3,999 for 6,000. Open source is free with unlimited usage on public repositories, but that does not provide unlimited private-repository usage.

The listed add-on rates can use a budget quickly: PR scanning costs 1 credit per scan, Context Guardrails 1 per context item, Dependency Validation 5 per private investigation, and Red Teaming 500 per report. One red-team report therefore uses all 500 credits in the Starter plan. Security-One-27B is priced separately at $0.05 per million input tokens.

Annual paid plans are billed upfront and add 20% credits, but annual prices are not published. The published pricing information also does not specify how many context items, investigations, or reports a typical team needs, or publish a free-trial offer. Buyers will need to estimate usage before deciding whether the monthly allowance fits.

Superagent has two free options: Open source includes unlimited use on public repositories, while Free adds 100 private credits per month. Paid plans start at $499 per month for 500 credits. Growth is $1,499 for 2,000 credits, and Scale is $3,999 for 6,000. That’s a $1,000 jump from Starter to Growth, then $2,500 from Growth to Scale; seats are unlimited on every listed plan.

Credits can go quickly: PR scanning costs 1 credit per scan, private dependency investigations cost 5, and a red-team report costs 500. Context Guardrails use 1 credit per item checked. Security-One-27B is priced separately at $0.05 per 1 million input tokens. Public dependency investigations are free.

The pricing page lists monthly prices. Annual paid plans are billed upfront and add 20% credits, but annual prices are not published. The free trial terms are also not published, so confirm the billing details before budgeting.

Superagent is a security layer, not an agent builder, so switching is mainly about fitting its checks into your existing development and agent workflows. The product description says findings can go into existing workflows, and the plans include API and MCP access. The published information does not specify migration steps, supported integrations, or how to export findings and configuration. Confirm those details before moving an established process.

Credit limits may be the bigger switching constraint than seats: private scans and red-team reports consume credits, while the free plan has 100 private credits per month. Estimate your expected scan volume against the plan caps before replacing another tool. Public-repository teams may have less cost pressure because the Open source plan includes unlimited public-repository usage.

We don’t have Reddit posts or comments in the published sources, so we can’t verify why users leave or what migration problems they report. Data export, contract terms, and offboarding requirements are also not published here. Get those answers from Superagent before you depend on it for security records.