# superagent-ai/superagent

Superagent protects your AI applications against prompt injections, data leaks, and harmful outputs. Embed safety directly into your app and prove compliance to your customers.

Repository: https://github.com/superagent-ai/superagent
Canonical: https://ross.abutalabs.com/products/superagent-ai-superagent
Homepage: https://superagent.sh
Language: TypeScript
License: MIT
License Family: permissive
Topics: ai, llm, anthropic, openai, security, guardrails, prompt-injection
Last push: 2026-08-25T09:45:01+00:00

## Health v2 (maintenance only)
Score: 78/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 99, release rhythm 47, longevity 86
- inputs: {"age_days": 1211, "days_push": 8, "days_rel": 353, "gap_med": 0.0, "n_releases_24m": 19}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 6719, forks 964 (observed 2026-08-28T04:09:48.040096+00:00)

## What it is
Superagent is an open-source SDK and platform for securing AI applications and agents, offering runtime guardrails that block prompt injections, PII/secret redaction, repository threat scanning, and red-team testing. It integrates via TypeScript and Python SDKs, GitHub Apps for PR security checks, and MCP tools for agent-facing context scoring.

## Use cases
- block prompt injection attacks in my llm app
- redact pii and secrets from user input before sending to an llm
- scan a github repo for malicious instructions targeting ai agents
- run red team tests against my ai agent
- add guardrails to what my coding agent can do at runtime
- score packages and urls before my agent consumes them
- add security checks to pull requests on github

## When to choose
- you are building LLM or agent applications and need runtime protection against prompt injection and data leaks
- you want automated security scanning and red teaming for AI-native codebases via GitHub
- you need compliance-friendly PII/PHI redaction in AI pipelines
- you want to gate what agents consume (files, URLs, packages, MCP servers) with trust scores

## When to avoid
- you need general-purpose application security testing unrelated to AI/LLM workloads
- you require fully offline, self-contained guardrails with no cloud API dependency
- your stack has no TypeScript or Python SDK support and you cannot call a REST API

## Facets
- artifact type: library
- maturity: active
- function: security, llm-inference, agent-framework, mcp, vulnerability-scanning, sdk, middleware
- domain: security, artificial-intelligence, large-language-models, developer-tools, privacy
- platform: python, cloud, self-hosted
- tags: guardrails, prompt-injection, pii-redaction, red-teaming, supply-chain-security, ai-safety, llm-security, runtime-protection, ai-agents, nodejs, typescript, web-server

## Member repositories
- superagent-ai/superagent (main) score 78

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:09:48.040096+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:42:55.212887+00:00, confidence not recorded.
  - readme: https://github.com/superagent-ai/superagent (fetched 2026-08-28T04:09:48.040096+00:00, sha 628f0c16ccb8)
  - homepage: https://superagent.sh (fetched 2026-08-29T08:39:17.737129+00:00, sha 0aea52327313)
  - site_page: https://www.superagent.sh/docs (fetched 2026-08-29T08:39:17.746641+00:00, sha db6e2058a3cc)
  - site_page: https://www.superagent.sh/docs/get-started/install-github-apps (fetched 2026-08-29T08:39:17.751550+00:00, sha effdff591c7d)
  - site_page: https://www.superagent.sh/about (fetched 2026-08-29T08:39:17.753564+00:00, sha 198d4cddac94)
  - site_page: https://www.superagent.sh/changelog (fetched 2026-08-29T08:39:17.748878+00:00, sha af8bb8ba32b3)
- Data as of 2026-08-30T08:39:29.467469+00:00.