Moltline Humanizer

Official5 tools

by GarphenGate · Python

Find AI-isms with evidence and fingerprint a writing voice from samples. 3 of 5 free.

com.moltlinestudio/humanizer MCP Server

This MCP server—com.moltlinestudio/humanizer—helps “find AI-isms with evidence” and “fingerprint a writing voice from samples.” It is one of Moltline Studio’s hosted MCP servers, accessed via a thin stdio bridge that proxies JSON-RPC messages between an MCP client and the hosted backend.

🛠️ Key Features

Finds AI-isms with evidence
Fingerprints a writing voice from samples
5 total tools
Topics include: agent-skills, ai-agents, claude, mcp, mcp-server, model-context-protocol

🚀 Use Cases

Detect AI-like phrasing using evidence
Identify or characterize a writing voice from provided samples

⚡ Developer Benefits

Thin, dependency-free stdio bridge
Proxies newline-delimited JSON-RPC “verbatim”
No telemetry; no dependencies beyond Python 3.9+ standard library

⚠️ Limitations

3 of 5 tools are free

Topics

agent-skillsai-agentsclaudemcpmcp-servermodel-context-protocol

Captured live from the server via tools/list.

ai_tell_scan

Scan a draft for the measurable tells of AI-generated prose. FREE. Flags stock phrases (with exact quotes), structural reflexes, uniform sentence rhythm, em-dash overuse, and hedging boilerplate — every flag cites the actual text. Typical input {"text": "<draft>"} returns {"reads_human_score": 0-100, "metrics": {"burstiness": ..., "avg_sentence_len": ..., ...}, "evidence": [{"type": "stock_phrase", "quote": "..."}], "note": "..."}. Use for a first read on whether a draft carries machine-writing signals. Reports measurable patterns, not a verdict on who wrote the text, and must not be used to accuse a person of anything. Not for rhythm detail (burstiness_report) or for a rewrite brief (humanize_plan). Errors: on invalid, missing, or malformed input this tool never raises a protocol error — it returns {"error": "<what is wrong and how to fix it>"} (for example {"error": "empty text"}). Every call is read-only and idempotent, so after correcting the input it is always safe to retry.

Parameters1

text

string

required

The draft to scan — at least one full sentence of plain text.

Raw schema

{
  "type": "object",
  "properties": {
    "text": {
      "type": "string",
      "description": "The draft to scan — at least one full sentence of plain text."
    }
  },
  "required": [
    "text"
  ],
  "additionalProperties": false
}

voice_fingerprint

Build a measurable voice profile from samples of a person's real writing. FREE. Feed it 2+ samples (emails, posts, essays — 150+ words total) and use the result with humanize_plan / verify_rewrite. Typical input {"samples": ["<email text>", "<blog post>"]} returns {"label": "my-voice", "target_metrics": {"avg_sentence_len": ..., "burstiness": ..., ...}, "favorite_words": [...], "signature_habits": ["..."], "words_analyzed": N}. Use on samples the person actually wrote, to build a target profile. Not for scoring an unknown draft (ai_tell_scan) and not on text the person did not write. Errors: on invalid, missing, or malformed input this tool never raises a protocol error — it returns {"error": "<what is wrong and how to fix it>"} (for example {"error": "need 150+ words of real writing across the samples"}). Every call is read-only and idempotent, so after correcting the input it is always safe to retry.

Parameters2

samples	array	required	List of real writing samples by the same person; 150+ words combined across all samples.
label	string	optional	Optional name for this voice profile. Default "my-voice".

Raw schema

{
  "type": "object",
  "properties": {
    "samples": {
      "items": {
        "type": "string"
      },
      "type": "array",
      "description": "List of real writing samples by the same person; 150+\nwords combined across all samples."
    },
    "label": {
      "default": "my-voice",
      "type": "string",
      "description": "Optional name for this voice profile. Default \"my-voice\"."
    }
  },
  "required": [
    "samples"
  ],
  "additionalProperties": false
}

burstiness_report

Map a draft's sentence rhythm and find where it goes flat. FREE. Reports per-sentence lengths, the burstiness coefficient, and runs of similar-length sentences. Typical input {"text": "<draft>"} returns {"sentence_lengths": [12, 14, 13, 5, 28], "burstiness": 0.52, "flat_runs": [{"sentences": "1-3", "lengths": [12, 14, 13]}], "tip": "..."}. Use when prose reads flat and sentence-length pattern is the suspect. Not for a full inventory of tells (ai_tell_scan). Errors: on invalid, missing, or malformed input this tool never raises a protocol error — it returns {"error": "<what is wrong and how to fix it>"} (for example {"error": "need 3+ sentences"}). Every call is read-only and idempotent, so after correcting the input it is always safe to retry.

Parameters1

text

string

required

The draft to analyze; needs at least 3 sentences.

Raw schema

{
  "type": "object",
  "properties": {
    "text": {
      "type": "string",
      "description": "The draft to analyze; needs at least 3 sentences."
    }
  },
  "required": [
    "text"
  ],
  "additionalProperties": false
}

humanize_plan

Produce a precise rewrite brief that de-AIs a draft, with numeric targets. PREMIUM (license). Lists every flagged tell with its fix and sentence-rhythm surgery targets; when a voice_fingerprint result is supplied, adds numeric targets to hit that person's voice. Apply the brief with your agent, then confirm with verify_rewrite. Typical input {"text": "<draft>", "fingerprint": <voice_fingerprint result>} returns {"current_score": 0-100, "edit_list": ["..."], "numeric_targets": {...}, "process": ..., "integrity_note": ...}. Use after a scan has identified what to fix; returns a brief, not rewritten prose. Not for checking whether a rewrite worked (verify_rewrite). Errors: on invalid, missing, or malformed input this tool never raises a protocol error — it returns {"error": "<what is wrong and how to fix it>"}. Every call is read-only and idempotent, so after correcting the input it is always safe to retry.

Parameters2

text	string	required	The AI draft to plan a rewrite for.
fingerprint	object	optional	Optional voice profile object exactly as returned by voice_fingerprint; omit for a style-only pass.

Raw schema

{
  "type": "object",
  "properties": {
    "text": {
      "type": "string",
      "description": "The AI draft to plan a rewrite for."
    },
    "fingerprint": {
      "additionalProperties": true,
      "default": null,
      "type": "object",
      "description": "Optional voice profile object exactly as returned by\nvoice_fingerprint; omit for a style-only pass."
    }
  },
  "required": [
    "text"
  ],
  "additionalProperties": false
}

verify_rewrite

Verify a rewrite actually improved: score delta, meaning check, voice distance. PREMIUM (license). Compares reads-human score before/after, a meaning-preservation proxy, and (with a fingerprint) numeric distance to the target voice. Typical input {"original": "<draft>", "rewrite": "<edited draft>"} returns {"score_before": N, "score_after": N, "score_delta": N, "content_word_retention_pct": N, "remaining_tells": [...], "verdict": "Improved — ship it" | "Marginal — ..."}. Use only when both the before and the after text are available. Not for scoring a single draft (ai_tell_scan). Errors: on invalid, missing, or malformed input this tool never raises a protocol error — it returns {"error": "<what is wrong and how to fix it>"} (for example {"error": "both texts must be non-empty"}). Every call is read-only and idempotent, so after correcting the input it is always safe to retry.

Parameters3

original	string	required	The draft before editing.
rewrite	string	required	The same content after the humanize_plan edits.
fingerprint	object	optional	Optional voice profile object exactly as returned by voice_fingerprint, to measure distance to the target voice.

Raw schema

{
  "type": "object",
  "properties": {
    "original": {
      "type": "string",
      "description": "The draft before editing."
    },
    "rewrite": {
      "type": "string",
      "description": "The same content after the humanize_plan edits."
    },
    "fingerprint": {
      "additionalProperties": true,
      "default": null,
      "type": "object",
      "description": "Optional voice profile object exactly as returned by\nvoice_fingerprint, to measure distance to the target voice."
    }
  },
  "required": [
    "original",
    "rewrite"
  ],
  "additionalProperties": false
}

README.md9 KBView on GitHub

moltline-mcp

A thin, dependency-free stdio bridge to Moltline Studio's fleet of 22 hosted MCP servers.

Most modern MCP clients can connect to the fleet directly over Streamable HTTP — no install needed (see Direct connection below). This bridge exists for clients that only speak the stdio transport: it proxies newline-delimited JSON-RPC between your client and the hosted server, verbatim, with no telemetry and no dependencies beyond the Python 3.9+ standard library.

160 tools across 22 servers; 110 are free.
Free tier: no registration, no account, no credentials — connect and call tools immediately.
Premium tools are unlocked with a Moltline license. Set MOLTLINE_LICENSE in your MCP client's environment; the bridge sends it as the X-Moltline-License header and never writes it to argv, URLs or logs. See auth.md.
Independently checked by MCPize on the direct endpoints (scans of 5–6 September 2026: 6 A+, 15 A, 1 B+; scores 89–98). All 22 servers score 100 on clarity, 100 on reliability and 100 on trust, 95 on protocol and 87 on safety; the one finding raised on every server is that the free tier accepts anonymous access — which is the point of it — and the spread between servers sits in MCPize's "Market" dimension (a crowded category). Each server's MCPize listing links its own public result.

Find it on: Smithery · MCPize · Glama · Clawmart · MCP Registry

The fleet

Server	Tools	What it does
`catalog`	10	Search 138 agent skills and personas by plain-language job; preview any product.
`codereview`	7	Risk-scan a diff, flag AI-generated-code tells, find secrets, report complexity.
`govern`	8	Audit an MCP config or SKILL.md for over-broad scope and prompt-injection risk.
`timeops`	5	Business days, meeting overlap, recurrence expansion, deadline and SLA math.
`data`	7	Paste-your-data analytics: CSV profiling, A/B tests, correlation, growth, cohorts.
`humanizer`	5	Find AI-isms with evidence, fingerprint a writing voice, measure burstiness.
`business`	8	30 finance, bookkeeping, legal-ops and SMB operations persona-skill products.
`creator`	8	20 blogging, brand-voice, copywriting, video and social persona-skill products.
`educator`	7	8 curriculum, classroom, accommodations and exam-prep persona-skill products.
`outbound`	8	7 outreach, sequencing, deliverability, call-coaching and CRM-hygiene products.
`personal`	9	20 inbox, calendar, travel, meals and family-logistics persona-skill products.
`research`	8	7 research navigation, thesis, note-taking and citation persona-skill products.
`skillmd-lint`	6	Lint a SKILL.md for frontmatter, structure, secrets and size. All tools free.
`merchant`	6	Commerce arithmetic: processor fees, charge-to-net, invoice totals, proration.
`shipping`	6	Dimensional weight, parcel fit, landed cost, freight class, rate cards.
`shopify`	6	Check a product CSV, URL handles, variant matrices and metafield keys.
`dropship`	6	Dropshipping unit economics: margin, lead time, SKU mapping, price ladders.
`recall`	6	A portable knowledge-graph memory you pass in and get back. No database.
`vision`	6	Image header probing, bbox conversion, resize plans, color and detection math.
`regclock`	9	Incident-reporting deadlines from the legal text (EU CRA, NIS2, DORA, GDPR, UK GDPR, HIPAA, SEC 8-K), cited.
`taxlots`	8	Crypto cost basis: parse exchange CSVs, lint, match transfers, FIFO/LIFO/HIFO lots, Form 8949 rows, 1099-DA diff.
`optimize`	11	OR-Tools routing, 3-D packing, 1-D/2-D cutting, knapsack and shift rostering with stated limits.

Machine-readable discovery: api-catalog · per-server card at https://mcp.moltlinestudio.com/<server>/.well-known/mcp/server-card.json · fleet health.

Direct connection (preferred)

If your MCP client supports Streamable HTTP (Claude Desktop, Claude Code, Cursor, and most current clients do), point it straight at the hosted endpoint — nothing to install:

json

{
  "mcpServers": {
    "moltline-timeops": {
      "type": "http",
      "url": "https://mcp.moltlinestudio.com/timeops"
    }
  }
}

Swap timeops for any slug in the table above.

The stdio bridge (for stdio-only clients)

Install

Run directly (recommended) — the bridge is a single stdlib-only file, so a clone is all you need:

bash

git clone https://github.com/GarphenGate/moltline-mcp.git
python3 moltline-mcp/moltline_mcp.py timeops   # any of the 22 slugs; default: catalog

Docker — see Docker below if you prefer a container.

A PyPI package is planned; until then, use the methods above.

Client configuration

Point your MCP client at the script with an absolute path:

json

{
  "mcpServers": {
    "moltline-timeops": {
      "command": "python3",
      "args": ["/absolute/path/to/moltline-mcp/moltline_mcp.py", "timeops"]
    }
  }
}

Or run it through Docker (after docker build -t moltline-mcp .):

json

{
  "mcpServers": {
    "moltline-timeops": {
      "command": "docker",
      "args": ["run", "-i", "--rm", "moltline-mcp", "timeops"]
    }
  }
}

CLI

text

python3 moltline_mcp.py [server] [--timeout SECONDS] [--list] [--version]

server — one of the 22 slugs (default catalog); anything else fails fast with the valid list.
--timeout — per-request timeout in seconds (default 300, or env MOLTLINE_TIMEOUT).
--list — print the server slugs and exit.
Env MOLTLINE_BASE_URL — override the fleet base URL (for testing).
Env MOLTLINE_LICENSE — your Moltline license, sent as the X-Moltline-License header to unlock premium tools. Omit it and the bridge runs free-tier only.
Env MOLTLINE_DEBUG=1 — diagnostic logging on stderr (stdout stays protocol-clean). The license is never logged; debug output reports only whether one is configured.

Docker

The bridge also runs containerized (stdio in, network egress to the fleet required):

bash

docker build -t moltline-mcp .
docker run -i --rm moltline-mcp timeops   # any of the 22 slugs; default: catalog

See Client configuration for the matching mcpServers entry.

How it works

One process per server connection. Each JSON-RPC message read from stdin is POSTed to https://mcp.moltlinestudio.com/<server>; JSON and SSE-framed responses are relayed back to stdout as newline-delimited JSON. The bridge tracks the server-assigned Mcp-Session-Id and echoes the negotiated MCP-Protocol-Version, per the Streamable HTTP transport spec. It never inspects, stores, or reports your traffic.

Tests

The bridge ships with an offline test suite - no network, no dependencies:

bash

python -m unittest -v

Twenty-two tests cover the parts that actually break in a transport bridge: newline-delimited framing, SSE event splitting (multiple events, multi-line data: payloads, a trailing event with no blank line, comments and named events), session-id capture and reuse, protocol-version negotiation on initialize only, and error relay - a JSON-RPC error document is passed through verbatim, anything else becomes a well-formed -32603. CI runs them on Python 3.9, 3.12 and 3.13, plus a Docker build and container smoke test.

Access and licensing

Free tier — anonymous. No registration or credentials; all free tools work immediately.
Premium — unlocked with a Moltline license purchased (human-in-the-loop) at moltlinestudio.com. Set MOLTLINE_LICENSE in the environment and the bridge forwards it as the X-Moltline-License request header. It is deliberately not a command-line flag: argv is world-readable through ps. It is never placed in a URL and never logged. Details: auth.md.

License

Install

Tools (5)

ai_tell_scanvoice_fingerprintburstiness_reporthumanize_planverify_rewrite

Remote endpoint

Streamable HTTP

Hosted server - connect over the network, no local install.

url

https://mcp.moltlinestudio.com/humanizer

Add to VS Code Add to Cursor

claude_desktop_config.json

json

{
  "mcpServers": {
    "humanizer": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://mcp.moltlinestudio.com/humanizer"
      ]
    }
  }
}

Desktop config is stdio-only; this bridges via mcp-remote. Native remote: Settings > Connectors.

Moltline Humanizer

com.moltlinestudio/humanizer MCP Server

🛠️ Key Features

🚀 Use Cases

⚡ Developer Benefits

⚠️ Limitations

Topics

ai_tell_scan

voice_fingerprint

burstiness_report

humanize_plan

verify_rewrite

moltline-mcp

The fleet

Direct connection (preferred)

The stdio bridge (for stdio-only clients)

Install

Client configuration

CLI

Docker

How it works

Tests

Access and licensing

Links

License

Install

Remote endpoint

Related servers