mixlayer

Mixlayer CLI

Privacy-first Mixlayer CLI for OpenAI-compatible chat, local PII redaction and vaulting, reasoning-aware model ladders, live model catalog refresh, and ledger-backed cost proofs.

Created by @cathrynlavery (Cathryn Lavery).

What This CLI Adds

Mixlayer exposes one OpenAI-compatible chat endpoint, but its real CLI value is the model ladder, reasoning_content, seed determinism, and the free 4B tier. This CLI keeps the generated chat and models wrappers and adds a local intelligence layer:

• shield masks sensitive data before frontier-model calls, writes privacy receipts, and keeps the reversible token vault local.
• ladder, escalate, and council compare default model rungs, reason traces, latency, and cost. Pass explicit Mixlayer model IDs when you want a different set.
• ask, replay, grep, models query, and sql turn prompts, answers, reasoning, and model metadata into a local SQLite ledger.
• savings and compare calculate cost deltas from local ledger evidence rather than vendor claims.

Install

The recommended path installs both the mixlayer-pp-cli binary and the pp-mixlayer agent skill (Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, and other agents supported by the upstream skills CLI) in one shot:

npx -y @mvanhorn/printing-press-library install mixlayer

For CLI only (no skill):

npx -y @mvanhorn/printing-press-library install mixlayer --cli-only

For skill only — installs the skill into the same agents as the default command above, but skips the CLI binary (use this to update or reinstall just the skill):

npx -y @mvanhorn/printing-press-library install mixlayer --skill-only

To constrain the skill install to one or more specific agents (repeatable — agent names match the skills CLI):

npx -y @mvanhorn/printing-press-library install mixlayer --agent claude-code
npx -y @mvanhorn/printing-press-library install mixlayer --agent claude-code --agent codex

Without Node (Go fallback)

If npx isn't available (no Node, offline), install the CLI directly via Go (requires Go 1.26.4 or newer):

go install github.com/mvanhorn/printing-press-library/library/ai/mixlayer/cmd/mixlayer-pp-cli@latest

This installs the CLI only — no skill.

Pre-built binary

Download a pre-built binary for your platform from the latest release. On macOS, clear the Gatekeeper quarantine: xattr -d com.apple.quarantine <binary>. On Unix, mark it executable: chmod +x <binary>.

Install for Hermes

Install the CLI binary first. The installer writes binaries to a per-user managed bin directory by default: $HOME/.local/bin on macOS/Linux and %LOCALAPPDATA%\Programs\PrintingPress\bin on Windows.

npx -y @mvanhorn/printing-press-library install mixlayer --cli-only

Then install the focused Hermes skill.

From the Hermes CLI:

hermes skills install mvanhorn/printing-press-library/cli-skills/pp-mixlayer --force

Inside a Hermes chat session:

/skills install mvanhorn/printing-press-library/cli-skills/pp-mixlayer --force

Restart the Hermes session or gateway if the newly installed skill is not visible immediately.

Install for OpenClaw

Install both the CLI binary and the focused OpenClaw skill. The installer defaults binaries to a per-user bin directory ($HOME/.local/bin on macOS/Linux, %LOCALAPPDATA%\Programs\PrintingPress\bin on Windows):

npx -y @mvanhorn/printing-press-library install mixlayer --agent openclaw

Restart the OpenClaw session or gateway if the newly installed skill is not visible immediately.

Use with Claude Desktop

This CLI ships an MCPB bundle — Claude Desktop's standard format for one-click MCP extension installs (no JSON config required).

To install:

1. Download the .mcpb for your platform from the latest release.
2. Double-click the .mcpb file. Claude Desktop opens and walks you through the install.
3. Fill in MIXLAYER_API_KEY when Claude Desktop prompts you.

Requires Claude Desktop 1.0.0 or later. Pre-built bundles ship for macOS Apple Silicon (darwin-arm64) and Windows (amd64, arm64); for other platforms, use the manual config below.

Manual JSON config (advanced)

If you can't use the MCPB bundle (older Claude Desktop, unsupported platform), install the MCP binary and configure it manually.

go install github.com/mvanhorn/printing-press-library/library/ai/mixlayer/cmd/mixlayer-pp-mcp@latest

Add to your Claude Desktop config (~/Library/Application Support/Claude/claude_desktop_config.json):

{
  "mcpServers": {
    "mixlayer": {
      "command": "mixlayer-pp-mcp",
      "env": {
        "MIXLAYER_API_KEY": "<your-key>"
      }
    }
  }
}

Quick Start

1. Install

See Install above.

2. Set Up Credentials

Get your access token from your API provider's developer portal, then store it:

mixlayer-pp-cli auth set-token YOUR_TOKEN_HERE

Or set it via environment variable:

export MIXLAYER_API_KEY="your-token-here"

3. Verify Setup

mixlayer-pp-cli doctor

This checks your configuration and credentials.

4. Try Your First Command

mixlayer-pp-cli models list

Privacy Firewall

mixlayer-pp-cli shield scan customer-export.csv --max-risk 0 --json
mixlayer-pp-cli shield redact customer-export.csv --diff -o masked.csv
mixlayer-pp-cli shield ingest big-export.csv -o masked-corpus.csv --manifest tranches.json
mixlayer-pp-cli shield ask "What segments are at risk?" --data customer-export.csv --json
mixlayer-pp-cli vault list --json
mixlayer-pp-cli shield audit --json

shield ask redacts locally, re-runs detectors on the outbound masked payload, sends only the masked corpus, then rehydrates the answer from the local vault. The audit receipt records the outbound payload hash, model, guard-model label, masked entity count, and residual leak count; the tripwire itself is enforced by local detectors.

Model Ladder And Reasoning Ledger

mixlayer-pp-cli ask "Draft a migration plan" --model qwen/qwen3.5-27b --show-thinking --json
mixlayer-pp-cli replay run_abc123 --json
mixlayer-pp-cli grep "migration" --json
mixlayer-pp-cli ladder "Which option is cheapest?" --reasoning --json
mixlayer-pp-cli escalate "Classify these support tickets" --confidence 0.85 --json
mixlayer-pp-cli council "Pick the safest launch plan" --json
mixlayer-pp-cli models query "ctx>=128k tools reasoning" --json
mixlayer-pp-cli models query "kimi" --refresh --json
mixlayer-pp-cli sql "select model, count(*) from runs group by model" --json

Use mixlayer-pp-cli models list --json to ask Mixlayer for the live model catalog available to your API key. models query seeds a local cache with the documented Qwen rungs plus console-visible IDs such as qwen/qwen3.6-27b, qwen/qwen3.6-35b-a3b, and moonshotai/kimi-k2.7-code; add --refresh to merge the live /models response into that cache. --model, --rungs, --members, and --judge accept any Mixlayer model ID the API authorizes.

Cost Proof

mixlayer-pp-cli savings --vs gpt-frontier --json
mixlayer-pp-cli compare "Summarize this incident" --cheap qwen/qwen3.5-9b --json

Unique Features

These capabilities aren't available in any other tool for this API.

• shield ingest — Split large CSV/text inputs into deterministic tranches, redact each tranche through one shared local vault, and reassemble a masked corpus with a manifest.
• shield ask — Redact local data, run an outbound PII tripwire, send only the masked payload to a frontier model, rehydrate the answer locally, and save a privacy receipt.
• shield scan — Detect high-risk PII locally with CI-friendly exit codes and no upstream model call.
• shield redact — Write a masked copy of a file and populate the local token-to-value vault with stable pseudonyms.
• shield restructure — Drop columns, bucket numeric values, and coarsen dates to reduce re-identification risk before prompting.
• vault — List, rehydrate, rotate, or purge the local reversible pseudonym vault without sending values upstream.
• shield audit — Inspect outbound payload hashes, byte counts, model IDs, masked entity counts, and residual leak counts for shielded sends.
• ladder — Run one prompt across selected Mixlayer model rungs, optionally comparing reasoning_content, latency, token usage, and cost.
• escalate — Start on a cheap rung and climb only when the prior answer appears insufficiently confident.
• council — Fan out to member model rungs and ask a judge model to synthesize answers and flag disagreements.
• ask — Ask Mixlayer, optionally request reasoning_content, and save prompt, answer, model, seed, token usage, cost, and raw response locally.
• replay — Re-run a saved prompt with its original model and seed to check determinism or drift.
• grep — Search saved prompts, answers, and reasoning traces through the local FTS ledger.
• models query — Query a local model cache with DSL terms like ctx>=128k tools reasoning, seed it with documented and console-visible model IDs, and optionally refresh it from live /models.
• sql — Run SELECT-only SQLite queries over the local ledger, vault, audit, ladder, and model cache tables.
• savings — Roll up saved ledger cost against static GPT or Claude frontier baselines.
• compare — Run a prompt through a cheap rung and frontier rung, report the cost delta, and attach a local-history quality note.

Usage

Run mixlayer-pp-cli --help for the full command reference and flag list.

Paths & environment variables

This CLI separates local files into four path kinds:

Kind	Contents
config	User-editable settings such as config.toml and saved profiles
data	Durable local data: credentials.toml, data.db, cookies, browser-session proof files, and other auth sidecars
state	Runtime state such as persisted queries, jobs, and teach.log
cache	Regenerable HTTP/cache files

Each kind resolves independently. The ladder is:

1. Per-kind env var: MIXLAYER_CONFIG_DIR, MIXLAYER_DATA_DIR, MIXLAYER_STATE_DIR, or MIXLAYER_CACHE_DIR
2. --home <dir> for this invocation
3. MIXLAYER_HOME for a flat relocated root
4. XDG env vars: XDG_CONFIG_HOME, XDG_DATA_HOME, XDG_STATE_HOME, XDG_CACHE_HOME
5. Platform defaults matching existing installs

For containers and agent sandboxes, prefer a single relocated root:

export MIXLAYER_HOME=/srv/mixlayer
mixlayer-pp-cli doctor

Under MIXLAYER_HOME=/srv/mixlayer, the four dirs resolve to /srv/mixlayer/config, /srv/mixlayer/data, /srv/mixlayer/state, and /srv/mixlayer/cache.

MCP servers do not receive CLI flags from the host. Put relocation in the host env block:

{
  "mcpServers": {
    "mixlayer": {
      "command": "mixlayer-pp-mcp",
      "env": {
        "MIXLAYER_HOME": "/srv/mixlayer"
      }
    }
  }
}

Precedence matters in fleets: an ambient per-kind variable such as MIXLAYER_DATA_DIR overrides an explicit --home for that kind. Use MIXLAYER_HOME or the per-kind variables for durable fleet relocation; treat --home as the weaker per-invocation lever.

Relocation is one-way. Unsetting MIXLAYER_HOME does not move files back to platform defaults, and doctor cannot find credentials left under a former root. Move the files manually before unsetting relocation variables.

Existing installs keep working because the platform-default rung matches the legacy layout. On the first auth write, stored secrets leave config.toml and are consolidated into credentials.toml under the data directory. Run mixlayer-pp-cli doctor --fail-on warn to check path and credential-location warnings in automation.

Commands

chat

OpenAI-compatible chat completions

• mixlayer-pp-cli chat - Create a chat completion for the given model and messages

models

Browse the model catalog

• mixlayer-pp-cli models get - Retrieve a model by ID
• mixlayer-pp-cli models list - List all available models

Output Formats

# Human-readable table (default in terminal, JSON when piped)
mixlayer-pp-cli models list

# JSON for scripting and agents
mixlayer-pp-cli models list --json

# Filter to specific fields
mixlayer-pp-cli models list --json --select id,name,status

# Dry run — show the request without sending
mixlayer-pp-cli models list --dry-run

# Agent mode — JSON + compact + no prompts in one flag
mixlayer-pp-cli models list --agent

Agent Usage

This CLI is designed for AI agent consumption:

• Non-interactive - never prompts, every input is a flag
• Pipeable - --json output to stdout, errors to stderr
• Filterable - --select id,name returns only fields you need
• Previewable - --dry-run shows the request without sending
• Explicit retries - add --idempotent to create retries when a no-op success is acceptable
• Confirmable - --yes for explicit confirmation of destructive actions
• Piped input - write commands can accept structured input when their help lists --stdin
• Offline-friendly - sync/search commands can use the local SQLite store when available
• Agent-safe by default - no colors or formatting unless --human-friendly is set

Exit codes: 0 success, 2 usage error, 3 not found, 4 auth error, 5 API error, 7 rate limited, 10 config error.

Health Check

mixlayer-pp-cli doctor

Verifies configuration, credentials, and connectivity to the API.

Configuration

Run mixlayer-pp-cli doctor to see the resolved config, data, state, and cache directories. The platform-default config path is ~/.config/mixlayer-pp-cli/config.toml; --home, MIXLAYER_HOME, and per-kind env vars can relocate it.

Static request headers can be configured under headers; per-command header overrides take precedence.

Environment variables:

Name	Kind	Required	Description
MIXLAYER_API_KEY	per_call	Yes	Set to your API credential.

agentcookie (optional)

If you use agentcookie to sync secrets across machines, this CLI auto-adopts agentcookie-managed credentials with no extra setup. When the daemon writes to this CLI's config, mixlayer-pp-cli doctor reports agentcookie: detected and auth-status labels the source as agentcookie. Skip this section if you don't use agentcookie - the CLI works the same as any other.

Troubleshooting

Authentication errors (exit code 4)

• Run mixlayer-pp-cli doctor to check credentials
• Verify the environment variable is set: echo $MIXLAYER_API_KEY Not found errors (exit code 3)
• Check the resource ID is correct
• Run the list command to see available items

---

Generated by CLI Printing Press