Manifest — LLM Router & Observability for OpenClaw

Manifest is an OpenClaw plugin that:

• Routes every request to the most cost-effective model via a 23-dimension scoring algorithm (<2ms latency)
• Tracks costs and tokens in a real-time dashboard
• Sets limits with email alerts and hard spending caps

Source: github.com/mnfst/manifest — MIT licensed. Homepage: manifest.build

Security & Privacy

TL;DR — The plugin registers Manifest as a standard OpenAI-compatible provider and exposes three read-only agent tools. It does not export telemetry or make background network calls. When you select manifest/auto as your model, OpenClaw routes requests through the Manifest backend — the same way it routes to any other provider like Anthropic or OpenAI. In local mode, all data stays on your machine and no API key is needed.

What the plugin does

1. Registers a provider — adds manifest as an OpenAI-compatible provider with the auto model
2. Injects config — writes provider entry to ~/.openclaw/openclaw.json and auth profiles (standard plugin registration, reversed on uninstall)
3. Exposes 3 read-only tools — manifest_usage, manifest_costs, manifest_health (query your own usage data via the Manifest API)
4. Registers /manifest command — shows connection status

What the plugin does NOT do

• Does not export telemetry, traces, or metrics — the plugin has no OTLP or telemetry code
• Does not make background or periodic network calls — network calls happen only at startup (connection verification) and when agent tools are invoked by the user
• Does not change your default model — manifest/auto is added to the allowlist only, you must switch to it manually

How routing works

When you manually select manifest/auto as your model, OpenClaw sends requests to the Manifest backend's /v1/chat/completions endpoint — the same way it sends requests to any provider (Anthropic, OpenAI, etc.). The backend picks the optimal model based on conversation complexity. This is standard OpenClaw provider behavior, not a special plugin data flow.

Credential storage

• Cloud mode (manifest-model-router plugin): API key provided via openclaw providers setup manifest-model-router or MANIFEST_API_KEY env var. The key authenticates with the Manifest backend — standard provider auth.
• Local mode (manifest plugin): auto-generated key stored in ~/.openclaw/manifest/config.json with file mode 0600. No external service contacted.

Local mode

All data stays on your machine. The embedded server runs locally and no external calls are made.

Install Provenance

openclaw plugins install manifest installs the manifest npm package.

• Source: github.com/mnfst/manifest (packages/openclaw-plugins/manifest)
• License: MIT
• Author: MNFST Inc.

Verify before installing:

npm view manifest repository.url
npm view manifest dist.integrity

The package is published with npm provenance attestations. Verify with:

npm audit signatures

Setup (Local — recommended for evaluation)

No account or API key required. Dashboard data stays local; LLM requests still go to your configured providers.

openclaw plugins install manifest
openclaw gateway restart

Dashboard opens at http://127.0.0.1:2099. Data stored locally in ~/.openclaw/manifest/manifest.db. No account or API key needed.

To expose over Tailscale (requires Tailscale on both devices, only accessible within your Tailnet): tailscale serve --bg 2099

Setup (Cloud)

Two commands:

openclaw plugins install manifest-model-router
openclaw providers setup manifest-model-router
openclaw gateway restart

The setup wizard prompts for your API key from app.manifest.build → create an account → create an agent → copy the mnfst_* key. You can also set MANIFEST_API_KEY env var for CI/CD.

After restart, the plugin registers itself via standard OpenClaw plugin APIs:

• Adds manifest/auto to the model allowlist — your current default model is not changed
• Registers the manifest provider in ~/.openclaw/openclaw.json (reversed on uninstall)
• Exposes three read-only agent tools: manifest_usage, manifest_costs, manifest_health

Dashboard at app.manifest.build.

Verify connection

openclaw manifest

Shows: mode, endpoint reachability, auth validity, agent name.

Configuration Changes

On plugin registration, Manifest writes to these files using standard OpenClaw plugin APIs. All changes are reversed by openclaw plugins uninstall manifest:

No other files are modified. The plugin does not change your current default model.

What Manifest Answers

Manifest answers these questions about your OpenClaw agents — via the dashboard or directly in-conversation via agent tools:

Spending & budget

• How much have I spent today / this week / this month?
• What's my cost breakdown by model?
• Which model consumes the biggest share of my budget?
• Am I approaching my spending limit?

Token consumption

• How many tokens has my agent used (input vs. output)?
• What's my token trend compared to the previous period?
• How much cache am I reading vs. writing?

Activity & performance

• How many LLM calls has my agent made?
• How long do LLM calls take (latency)?
• Are there errors or rate limits occurring? What are the error messages?
• Which skills/tools are running and how often?

Routing intelligence

• What routing tier (simple/standard/complex/reasoning) was each request assigned?
• Why was a specific tier chosen?
• What model pricing is available across all providers?

Connectivity

• Is Manifest connected and healthy?

Agent Tools

Three read-only tools are available to the agent in-conversation:

Each accepts a period parameter: "today", "week", or "month".

All three tools are read-only — they query the agent's own usage data and never send message content.

LLM Routing

Routing only activates when you manually set your model to manifest/auto. When active, the Manifest backend scores each conversation across 23 dimensions and assigns one of 4 tiers:

Each tier maps to a model. Default models are auto-assigned per provider, but overridable in the dashboard under Routing.

Short-circuit rules:

• Messages <50 chars with no tools → Simple
• Formal logic keywords → Reasoning
• Tools present → floor at Standard
• Context >50k tokens → floor at Complex

Dashboard Pages

Supported Providers

Anthropic, OpenAI, Google Gemini, DeepSeek, xAI, Mistral AI, Qwen, MiniMax, Kimi, Amazon Nova, Z.ai, OpenRouter, Ollama. 300+ models total.

Uninstall

openclaw plugins uninstall manifest
openclaw gateway restart

This removes the plugin, provider config, and auth profiles. After uninstalling, manifest/auto is no longer available. If any agent uses it, switch to another model.

Troubleshooting

Auth errors in cloud mode: Verify the API key starts with mnfst_ and matches the key in the dashboard under Settings → Agent setup.

Port conflict in local mode: If port 2099 is busy, the plugin checks if the existing process is Manifest and reuses it. To change the port: openclaw config set plugins.entries.manifest.config.port <PORT>.

Plugin conflicts: Manifest conflicts with the built-in diagnostics-otel plugin. Disable it before enabling Manifest.

After backend restart: Always restart the gateway too (openclaw gateway restart).