Silmaril

Silmaril is a Y Combinator-backed runtime security company that builds an AI application firewall for agentic systems. The Silmaril Firewall classifies prompts, retrieved context, tool calls, tool responses, model output, and accumulated execution state before harmful outcomes materialize, returning a benign/malicious prediction plus a primary-outcome classification across secret exposure, information disclosure, control abuse, system compromise, and service disruption. It ships first-party TypeScript, Python, Go, and Java SDKs, LiteLLM and Vercel AI gateway guardrails, LangChain/LangGraph handlers, and agent plugins for Codex, Claude Code, OpenClaw, and Hermes, plus a hosted read-only MCP evidence server. Offered as managed and customer-controlled self-hosted container deployments.

Silmaril publishes 1 API on the APIs.io network. Tagged areas include Company, AI Security, AI Firewall, Prompt Injection, and Agent Security.

Silmaril’s developer surface includes documentation, API reference, getting-started guide, support, authentication, changelog, and 17 more developer resources.

🌐 Visit website 📡 Source on GitHub

32.7/100 thin ▬ flat Agent 29/100 agent aware Full breakdown ↓
scored 2026-07-27 · rubric v0.5

AccessSelf serve

1 APIs 1 MCP Servers

CompanyAI SecurityAI FirewallPrompt InjectionAgent SecurityRuntime SecurityGuardrailsMCPLLM SecurityDeveloper Tools

Kin Score

Kin Score How this is scored →
scored 2026-07-27 · rubric v0.5

Composite quality — 32.7/100 · thin

Contract Quality 0.0 / 25

Developer Ergonomics 13.0 / 20

Commercial Clarity 4.2 / 20

Operational Transparency 6.2 / 13

Governance 0.0 / 12

Discoverability 9.3 / 10

Agent readiness — 29/100 · agent aware

Machine-Readable Contract 0 / 18

Agentic Access Contract 0 / 15

MCP Server 12 / 12

Machine-Readable Auth 10 / 10

Idempotency 0 / 9

Stable Error Semantics 8 / 8

Request/Response Examples 0 / 7

Rate-Limit Signaling 0 / 7

Typed Event Surface 0 / 6

Agent Skills 0 / 5

Well-Known Catalog 0 / 4

Consent & Bot Identity 0 / 3

Improve this rating by publishing the missing artifacts — every area above can be raised, and the full rubric is at apis.io/rating/. This rating is computed from github.com/api-evangelist/silmaril: open an issue to ask a question, or submit a pull request to add artifacts. Want it done for you? Prioritized profiling — $2,500 →

APIs 1

Individual APIs this provider publishes, each with its own machine-readable definition.

Silmaril Firewall API

AI application firewall classify API. A Firewall client sends each sanitized classification input (user input, tool response, model output, or system prompt) with a hook label a...

MCP Servers 1

Model Context Protocol servers that expose these APIs to AI agents.

Get Started 2

Portal, sign-up, and the first successful call

DeveloperPortal

DeveloperPortal

GettingStarted

GettingStarted

Documentation 2

Reference material describing how the API behaves

Documentation

Documentation

APIReference

APIReference

Agent Surfaces 2

MCP servers, agent skills, and machine-readable catalogs

LLMsTxt

LLMsTxt

MCPServer

MCPServer

Design & Contract 5

Pagination, idempotency, versioning, errors, and events

Conventions

ErrorCatalog

Lifecycle

Conformance

DataModel

Build 3

SDKs, sample code, and the tooling you integrate with

GitHubOrganization

GitHubOrganization

Packages

Packages

SDKs

SDKs

Access & Security 4

Authentication, authorization, and security posture

Authentication

Authentication

Security

Security

VulnerabilityDisclosure

VulnerabilityDisclosure

DomainSecurity

DomainSecurity

Operate 3

Status, limits, changes, and where to get help

Support

Support

StatusPage

StatusPage

ChangeLog

ChangeLog

Commercial 2

Pricing, plans, and the legal terms of use

TermsOfService

TermsOfService

PrivacyPolicy

PrivacyPolicy

Source (apis.yml)

aid: silmaril
name: Silmaril
description: Silmaril is a Y Combinator-backed runtime security company that builds an AI application firewall for agentic
  systems. The Silmaril Firewall classifies prompts, retrieved context, tool calls, tool responses, model output, and accumulated
  execution state before harmful outcomes materialize, returning a benign/malicious prediction plus a primary-outcome classification
  across secret exposure, information disclosure, control abuse, system compromise, and service disruption. It ships first-party
  TypeScript, Python, Go, and Java SDKs, LiteLLM and Vercel AI gateway guardrails, LangChain/LangGraph handlers, and agent
  plugins for Codex, Claude Code, OpenClaw, and Hermes, plus a hosted read-only MCP evidence server. Offered as managed and
  customer-controlled self-hosted container deployments.
accessModel:
  pricing: unknown
  onboarding: self-serve
  trial: false
  try_now: false
  public: false
  label: Self-serve signup
  confidence: medium
  source:
  - authentication
  generated: '2026-07-22'
  method: derived
image: https://www.silmaril.dev/icon.svg
url: https://raw.githubusercontent.com/api-evangelist/silmaril/refs/heads/main/apis.yml
x-type: company
x-source: vc-portfolio
x-backed-by:
- y-combinator
x-tier: stub
x-tier-reason: portfolio-lead
x-enrichment:
  date: '2026-07-21'
  status: enriched
  artifacts_added: 13
  pass: local-v1
specificationVersion: '0.20'
created: '2026-07-17'
modified: '2026-07-21'
tags:
- Company
- AI Security
- AI Firewall
- Prompt Injection
- Agent Security
- Runtime Security
- Guardrails
- MCP
- LLM Security
- Developer Tools
apis:
- name: Silmaril Firewall API
  description: AI application firewall classify API. A Firewall client sends each sanitized classification input (user input,
    tool response, model output, or system prompt) with a hook label and optional tool name to a tenant-issued classify endpoint,
    and receives a prediction (BENIGN/MALICIOUS), a primary outcome, and score/threshold data. Credentials and the classify
    endpoint are issued by Silmaril (not self-serve). A companion hosted, read-only MCP evidence server exposes firewall findings
    to agents.
  humanURL: https://silmaril.dev/docs
  baseURL: https://firewall-mcp.silmaril.dev
  tags:
  - AI Firewall
  - Runtime Security
  - Prompt Injection
  properties:
  - type: Documentation
    url: https://silmaril.dev/docs
  - type: APIReference
    url: https://silmaril.dev/docs
common:
- type: DeveloperPortal
  url: https://silmaril.dev/docs
- type: Documentation
  url: https://silmaril.dev/docs
- type: APIReference
  url: https://silmaril.dev/docs
- type: GettingStarted
  url: https://silmaril.dev/docs
- type: GitHubOrganization
  url: https://github.com/Silmaril-Security
- type: Support
  url: https://cal.com/silmaril/30min
- type: StatusPage
  url: https://silmaril.dev/status
- type: TermsOfService
  url: https://silmaril.dev/terms
- type: PrivacyPolicy
  url: https://silmaril.dev/privacy
- type: LLMsTxt
  url: llms/silmaril-llms.txt
- type: Packages
  url: packages/silmaril-packages.yml
- type: SDKs
  url: packages/silmaril-packages.yml
- type: MCPServer
  url: mcp/silmaril-mcp.yml
- type: Authentication
  url: authentication/silmaril-authentication.yml
- type: Conventions
  url: conventions/silmaril-conventions.yml
- type: ErrorCatalog
  url: errors/silmaril-error-codes.yml
- type: ChangeLog
  url: changelog/silmaril-changelog.yml
- type: Lifecycle
  url: lifecycle/silmaril-lifecycle.yml
- type: Conformance
  url: conformance/silmaril-conformance.yml
- type: DataModel
  url: data-model/silmaril-data-model.yml
- type: Security
  url: security/silmaril-vulnerability-disclosure.yml
- type: VulnerabilityDisclosure
  url: security/silmaril-vulnerability-disclosure.yml
- type: DomainSecurity
  url: security/silmaril-domain-security.yml
maintainers:
- FN: Kin Lane
  email: kin@apievangelist.com
- FN: APIs.json
  email: info@apis.io