Agent Economy ReportAgent Economy ReportCAUTION version 1.0.0 · audited 2026-09-17
This package is a markdown-only writing-editing skill for an AI agent. It teaches the agent to find common AI-style wording and rewrite the text so it sounds more natural, without adding any executable program code. It requests only file reading, file writing, and a user-question tool; no shell, network, credential, or payment access is requested. The files contain documentation links to Wikipedia and ClawHub, but no instruction to call them at runtime. The publisher itself notes the skill could be used to make AI-generated writing appear human-written, so users should apply disclosure rules and review edited files.
| Declared purpose | Remove signs of AI-generated writing from text and rewrite it to sound more natural while preserving meaning and tone, based on Wikipedia's Signs of AI writing guide. |
| Observed behavior | The skill is an instruction set rather than executable code. It tells the agent to identify listed AI writing patterns, rewrite problematic sections, preserve meaning and tone, and output the humanized text. The frontmatter requests Read, Write, Edit, Grep, Glob, and AskUserQuestion. No runtime network, shell, credentials, payments, or persistence behavior is described. |
| Verdict, rules only | SAFE |
| Verdict, AI | CAUTION |
| Final verdict | CAUTION (never better than either pass) |
| capability | vs purpose | detail |
|---|---|---|
| filesystem_read | needed | Frontmatter permits Read, Grep, and Glob (SKILL.md lines 12, 15-16) so the agent can read input text or locate documents to humanize. |
| filesystem_write | needed | Frontmatter permits Write and Edit (SKILL.md lines 13-14) so the agent can output rewritten text or edit documents in place. |
| other | needed | Frontmatter permits AskUserQuestion (SKILL.md line 17) so the agent can ask the user about preferences or documents. |
| severity | finding |
|---|---|
| info | Declared purpose The frontmatter states the skill is for editing or reviewing text to remove AI-generated writing patterns. SKILL.md:4 Remove signs of AI-generated writing from text. Use when editing or reviewing |
| info | Requested capabilities are limited The allowed-tools section (SKILL.md lines 11-17) lists only Read, Write, Edit, Grep, Glob, and AskUserQuestion. It does not request shell, network, credential, or payment tools. SKILL.md:11 allowed-tools: |
| info | Agent role instruction This line assigns the agent a style-editing role. It is ordinary skill guidance and does not instruct access to credentials, payments, or the network. SKILL.md:22 You are a writing editor that identifies and removes signs of AI-generated text to make writing sound more natural and human. This guide is based on Wikipedia's |
| info | Installation command The README shows the package is installed through the ClawHub CLI. This is documentation for installing the skill, not a runtime shell or package-installation behavior inside the skill. README.md:10 clawdhub install humanizer |
| info | External reference link The README links to a Wikipedia guide. This is a documentation reference only; there is no instruction to fetch the URL or send data to it. README.md:25 Based on [Wikipedia's "Signs of AI writing"](https://en.wikipedia.org/wiki/Wikipedia:Signs_of_AI_writing) guide, maintained by WikiProject AI Cleanup. This comp |
| info | Publisher link The skill card links to a ClawHub publisher page. This is a documentation-only external link, not a runtime network destination. skill-card.md:9 [biostartechnology](https://clawhub.ai/user/biostartechnology) |
| low | Dual-use risk disclosed The publisher acknowledges the skill could hide AI authorship. The surrounding text says to use it only for legitimate style improvement and to follow disclosure rules. This is the main caution for users. skill-card.md:25 Risk: The skill can be used to make AI-generated writing appear human-written without appropriate disclosure. |
| info | Version mismatch The skill-card reports a metadata version mismatch: SKILL.md frontmatter says 2.1.1, while other metadata says 1.0.0. This is not malicious behavior, but the package metadata is inconsistent. skill-card.md:50 1.0.0 (source: server release metadata; artifact SKILL.md frontmatter says 2.1.1) |
0 AI finding(s) were dropped because their file, line or quote did not match the package.
| domain | service rating | where |
|---|---|---|
| clawhub.ai | not in the service index | skill-card.md:9 |
| en.wikipedia.org | not in the service index | README.md:25 |
| Binaries invoked | bash |
| Environment variables read | none |
| Hard-coded wallets | none |
| Pipes a download to a shell | 0 |
| eval / exec / subprocess | 0 |
| base64 blobs | 0 |
| File writes | 1 |
| Persistence | 0 |
| file | lines | sha256 |
|---|---|---|
README.md | 82 | 4f3f9414dc2b7605… |
SKILL.md | 437 | d7ce483d368e4301… |
_meta.json | 6 | 9371a63c2d97d9a1… |
skill-card.md | 54 | 3962ca08447d8182… |
For agents
JSON: https://agenteconomy.report/k/humanizer.audit.json · badge: https://agenteconomy.report/k/humanizer.audited.svg ·
skill rating: /k/humanizer · commission an audit of any skill: US$ 29 per version.
The complete published package of this exact version was downloaded from the registry and read statically; nothing was executed. A deterministic pass extracts network destinations, binaries, environment variables, writes, install commands, obfuscation markers, subprocesses and wallets, each with file and line. An AI then reads the whole package with those facts and writes the summary, the capabilities and the findings under a strict schema; every finding it produces must cite a file, a line and the exact text, or it is dropped. The final verdict is the worst of the two passes. The audit does not change the skill's trust tier (policy); the author may respond through the dispute channel and the response is published here. Commissioned by: the Agent Economy Report (free program: the 150 most downloaded skills, September 2026).