Get This Tool
Humanize
Pricing
- Model
- Free
Summary
Perplexity-based detectors like ZeroGPT flag your prose not because it reads like a robot, but because it lacks the statistical irregularity humans produce naturally — and most rewriting tools guess at fixes rather than model the underlying signal. Humanize is a pair of static skills, installable in one command, that apply research-grounded rules to close that gap.
The two skills — `humanize` and `ai-check` — work inside Claude Code, ChatGPT, Gemini, Codex CLI, Cursor, and comparable agents, not as a hosted API but as instruction files dropped into your agent's context. `humanize` rewrites text across nine documented levers drawn from 50+ peer-reviewed sources through April 2026. `ai-check` runs the reverse: forensic scoring with quoted evidence flagging the specific phrases that read as AI-generated. Because the skills are static files, there is no server, no rate limit, and no external dependency — but there is also no adaptive learning, no feedback loop, and no guarantee a detector updated after April 2026 won't develop new signals the rules don't cover yet.
Bottom line: Pick this when you need offline, auditable humanization rules you can drop into an existing LLM workflow — plan on revisiting the skill files manually the moment a detector shifts its scoring model in ways the April 2026 literature doesn't cover.
Community Performance Report Card
No community ratings yet. Be the first to rate this tool!
Community Benchmarks Community
Sign in to submit a benchmarkNo community benchmarks yet. Be the first to share a real-world data point.
Pros
Sign in to edit- Static, offline skill files with no external API dependency, so your text never leaves your local environment and there is no rate limit to hit during high-volume editing sessions.
- Nine documented humanization levers grounded in 50+ peer-reviewed sources, which means you can audit exactly why a rewrite changes a specific phrase rather than accepting a black-box output.
- `ai-check` quotes the specific phrases it flags as AI-generated, so you can target rewrites precisely instead of blanket-rewriting paragraphs that were already passing.
- Installs into Claude Code, Codex CLI, ChatGPT, Gemini, Cursor, Aider, OpenCode, Continue, and Copilot from a single command, so you embed the skill in whichever agent your team already uses rather than switching tools.
- MIT license with fully readable skill files, which means a team can fork, extend, or audit the rule set without negotiating access or reverse-engineering a closed system.
Cons
Sign in to edit- The rule set is frozen at the detection literature available through April 2026 — when a detector like ZeroGPT ships a model update that weights new signals, the skill files produce no defensive response until someone manually updates them, and there is no automated mechanism for that.
- There is no API, no CLI that accepts stdin, and no programmatic output format — bulk processing a document queue or integrating detection scores into a CI pipeline requires wrapping the skill invocation in agent automation that the project does not provide, and teams with that requirement switch to a hosted detection API instead.
- Benchmark results depend on which underlying LLM executes the skill and which model version it runs — the same skill file applied in GPT-4o versus a smaller local model produces different rewrite quality, and the project offers no normalization layer to account for that variance.
Community Reviews
Sign in to write a reviewNo reviews yet. Be the first to share your experience.
About
- Platforms
- Any LLM agent (Claude Code, Codex CLI, ChatGPT desktop, Gemini, Cursor, Aider, etc.)
- API Available
- No
- Self-Hosted
- Yes
- Last Updated
- 2026-07-08T13:04:52.664Z
Best For
Who it's for
- Users needing static humanization rules without APIs
- Forensic-style AI detection with cited evidence
- Integration into existing LLM agent workflows
- Research-grounded text processing
What it does well
- Rewriting AI-generated text to appear human-authored
- Forensic analysis of text for AI signals with evidence
- Avoiding perplexity-based detectors such as ZeroGPT
- Generating or editing content in Claude Code, ChatGPT, Gemini, or Codex CLI
Integrations
Discussion Community
Sign in to commentNo discussion yet. Sign in to start the conversation.
Compare Humanize
Spotted incorrect or missing data? Join our community of contributors.
Sign Up to ContributeCommunity Notes & Tips Community
Sign in to contributeBe the first to contribute. General notes, observations, gotchas, and tips from people who use this tool day-to-day.
Frequently Asked Questions
- Is Humanize free?
- Yes — Humanize is fully free to use. There is no paid tier.
- Is Humanize open source?
- Yes. Humanize is open source.
- Can I self-host Humanize?
- Yes. Humanize supports self-hosting on your own infrastructure.
- What platforms does Humanize support?
- Humanize is available on: Any LLM agent (Claude Code, Codex CLI, ChatGPT desktop, Gemini, Cursor, Aider, etc.).
Hours Saved & ROI Stories Community
Sign in to contributeBe the first to contribute. Concrete time/cost savings, with context. e.g. "Cut my code review backlog from 4h to 45m per week."
Curated lists that include this category
Humanize ships as two LLM skill files — `humanize` and `ai-check` — each containing static, research-grounded instructions that any compatible agent executes on demand. The core workflow is one install command that places the skills in the right directory for your agent, then a natural-language invocation: you point the skill at a piece of text, and the agent applies the rule set. No external API call leaves your environment. No account is required. The skills are MIT-licensed and readable in plain text, so every transformation rule is inspectable before you deploy.
The differentiating feature is the sourcing. Rather than heuristic rewrites or prompt-engineering folklore, the `humanize` skill models nine levers — syntactic variation, burstiness, lexical entropy, and related dimensions — drawn from detection literature spanning peer-reviewed sources through April 2026. The `ai-check` skill inverts this: it scores incoming text and quotes the specific phrases triggering AI-signal flags, giving you evidence rather than a black-box pass/fail score. The vendor-published benchmark describes 25 inputs evaluated against two independent scorers, though reproduction depends on the agent and model you run it in.
The tool fits teams that want humanization logic embedded in an existing LLM agent workflow without standing up a separate service, and researchers who need a citable, auditable rule set rather than a proprietary rewriter. It breaks when the target detector updates its scoring model after April 2026, when you need bulk processing through a REST endpoint, or when your pipeline requires structured JSON output with confidence scores — none of which the static skill architecture provides.
Installation paths differ per agent — the README documents separate commands for Claude Code, Codex CLI, ChatGPT desktop, Gemini, Cursor, Aider, OpenCode, Continue, and Copilot — but the skill content is identical across all of them. The `install.sh` script handles path resolution; the skills themselves are plain instruction files, which means updating them is a file replacement, not a dependency upgrade.