# mnfst/manifest

Connect Your Agents And Harnesses With Any Provider 🦚

Repository: https://github.com/mnfst/manifest
Canonical: https://ross.abutalabs.com/products/manifest
Homepage: https://manifest.build
Language: TypeScript
License: MIT
License Family: permissive
Topics: openclaw, cost-tracking, llm-observability, token-tracking, llm-router, hermes-agent, ai, ai-, byok, gateway, open-source, ai-sdk, openai-api, anthropic, subscription
Last push: 2026-08-25T22:53:42+00:00

## Health v2 (maintenance only)
Score: 94/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 99, release rhythm 84, longevity 100
- inputs: {"age_days": 1436, "days_push": 8, "days_rel": 28, "gap_med": 0, "n_releases_24m": 296}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 7481, forks 502 (observed 2026-08-28T04:10:00.205509+00:00)

## What it is
Manifest is an open-source LLM gateway and router that lets AI agents and apps reach all their model providers through one OpenAI- and Anthropic-compatible endpoint. It adds routing with fallbacks, spend limits, on-the-fly request repair (Autofix), and per-request cost and token observability, deployable via Docker or a hosted cloud.

## Use cases
- route agent LLM calls through one endpoint across multiple providers
- add fallbacks when a model or provider fails or is rate-limited
- track token usage and dollar cost per agent or harness
- use ChatGPT/Claude/Gemini subscriptions and BYOK keys behind one gateway
- self-host an LLM proxy with spend caps and budget alerts
- auto-repair malformed LLM requests that fail with 4xx errors
- connect local models and custom endpoints to agent harnesses

## When to choose
- you run AI agents that need reliable multi-provider LLM access with fallbacks
- you want self-hosted cost tracking and spend limits for LLM calls
- you want to pool subscriptions, API keys, and local models behind one OpenAI-compatible URL
- you need request logs and observability for every provider attempt

## When to avoid
- you need model training, fine-tuning, or hosting rather than routing
- you want a client-side SDK only without a gateway service
- you need a lightweight proxy with no dashboard or billing features
- you require team features like SSO on a free plan

## Facets
- artifact type: service
- maturity: active
- function: llm-inference, api-gateway, monitoring, rate-limiting, middleware
- domain: large-language-models, developer-tools, self-hosted, apis
- platform: self-hosted, cloud
- tags: llm-gateway, llm-router, autofix, byok, cost-tracking, token-tracking, openai-compatible, anthropic-compatible, fallback-routing, observability, ai-agents, docker, web-server, nodejs

## Member repositories
- mnfst/manifest (main) score 94

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:10:00.205509+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 2026-08-29T17:37:35.964619+00:00, confidence not recorded.
  - readme: https://github.com/mnfst/manifest (fetched 2026-08-28T04:10:00.205509+00:00, sha 62079904ce5c)
  - homepage: https://manifest.build (fetched 2026-08-29T08:33:19.851441+00:00, sha 8a940b54c632)
  - site_page: https://manifest.build/docs/llm-gateway (fetched 2026-08-29T08:33:19.861680+00:00, sha 3440618e7ebc)
  - site_page: https://manifest.build/docs/autofix (fetched 2026-08-29T08:33:19.864733+00:00, sha d1770dcec147)
  - site_page: https://manifest.build/docs/observability (fetched 2026-08-29T08:33:19.868501+00:00, sha 845ddbe9258e)
  - site_page: https://manifest.build/docs/introduction (fetched 2026-08-29T08:33:19.873842+00:00, sha 24828f3fefa9)
  - site_page: https://manifest.build/docs/self-hosted (fetched 2026-08-29T08:33:19.876006+00:00, sha 4f8fe815fa8f)
  - site_page: https://manifest.build/pricing (fetched 2026-08-29T08:33:19.871466+00:00, sha f5e24c41f5c2)
  - site_page: https://manifest.build/pricing?tab=cloud&modal=enterprise (fetched 2026-08-29T08:33:19.878549+00:00, sha a3c9e5bd2f23)
  - site_page: https://manifest.build/changelog (fetched 2026-08-29T08:33:19.880766+00:00, sha 34fede6c9829)
- Data as of 2026-08-30T08:39:29.467469+00:00.