dgent

De-agent your code. Run dgent to check files and commit messages for AI tells before committing — flags naming patterns, catch-rethrow, AI vocabulary, emoji,...

Security Scan

The skill is coherent with its stated purpose (a CLI tool that detects and strips 'AI tells' from code and commit messages); its requirements (a dgent binary installed from the @itscojac/dgent npm package) and runtime instructions match that purpose and are proportionate.

Purpose & Capability

Name/description, required binary (dgent), and declared npm install (@itscojac/dgent) align with the stated function of scanning/fixing files and commit messages for AI artifacts. No unrelated credentials, binaries, or config paths are requested.

Instruction Scope

SKILL.md only instructs the agent to run the dgent CLI (run, scan, rules) against files or commit messages and describes suppressions and workflow. It does not instruct reading unrelated system files, exfiltrating data, or accessing credentials. One note: the doc says dgent can run as git hooks, which implies repository-level modification of hooks when the package is installed or initialized—this is consistent with the tool's purpose but operators should be aware hooks may be added/modified.

Install Mechanism

Install is a single npm package (@itscojac/dgent) that provides the dgent binary. Using an npm package for a CLI is expected. No network downloads from arbitrary URLs or extract-from-URL steps are present in the skill metadata.

Credentials

No environment variables, credentials, or config paths are requested. The lack of broad permissions is proportionate to a lint/fix CLI tool.

Persistence & Privilege

Skill is not forced-always nor requesting elevated privileges. The possible repository-level persistence is that dgent may install git hooks (as advertised) when you install or enable it—this is reasonable for a pre-commit/commit-msg-style tool but is the main persistence-related effect to be mindful of.

Assessment

This skill is internally consistent with a CLI linter/fixer for removing 'AI tells'. Before installing: 1) Verify the npm package and GitHub repo (@itscojac/dgent) and the maintainer's reputation; 2) prefer installing it as a project devDependency (or inspect the package contents) rather than globally if you want limited scope; 3) review the rule catalog locally with dgent rules --json to see exactly what it will flag/fix; 4) be aware it may install or modify git hooks in your repository when set up—review any changes to .git/hooks or install scripts; 5) check lockfiles / package integrity (sha/lock) if you require supply-chain assurance. No environment variables or secrets are requested by the skill metadata.

dgent — de-agent your code

Clean AI tells from code and commit messages. Runs as git hooks automatically, or on-demand via CLI.

When to use

• Before committing code written by an AI agent
• When asked to "clean up" or "de-agent" output
• To check if code has AI naming patterns, catch-rethrow, section headers, or noise comments
• To check if a commit message has AI vocabulary, emoji, or trailers

Quick check a file

dgent run --json <file>

Returns JSON with clean, fixes[], flags[], and output (cleaned content). Exit codes: 0 = clean or fixes applied, 1 = flags found.

Quick check — exit code only

dgent run --check <file>

Silent. Exit 0 if clean or fixes applied, 1 if flags found. Use in pre-commit scripts.

Scan entire directory

dgent scan --json [dir]

Returns JSON with per-file results. Use before committing multi-file changes.

Fix a file in place

dgent run --fix <file>

Applies all deterministic fix rules (strip trailers, emoji, section headers) and writes back.

Check a commit message

echo "your message" | dgent run --json --commit-msg -

Stdin with short non-code input auto-detects as commit-msg. Returns fixes (emoji, trailers) and flags (AI vocabulary).

Get the full rule catalog

dgent rules --json

Returns every rule with name, phase, type, enabled status, and complete pattern lists. Use this to know exactly what to avoid.

Patterns to avoid

These trigger flags (from dgent rules --json):

Commit message words: enhance, streamline, comprehensive, utilize, leverage, facilitate, robust, optimize

Commit message phrases: "this commit", "this change", "in order to", "aims to", "is designed to"

Naming suffixes: Manager, Handler, Processor, Service, Factory, Builder, Validator, Controller, Orchestrator, Coordinator

Identifier length: over 40 characters

Catch-rethrow: catch (e) { console.error(e); throw e; } — either handle the error or let it propagate

Suppress a specific flag

// dgent-ignore flag-naming
class DataProcessor { ... }  // not flagged

Supports // dgent-ignore, // dgent-ignore-next-line, // dgent-ignore <rule1> <rule2>.

Workflow for agents

1. Write code
2. Run dgent run --json <file> on each modified file
3. Fix any flags (rename identifiers, remove catch-rethrow, etc.)
4. Run echo "commit message" | dgent run --json --commit-msg - on the message
5. Fix any message flags (remove AI vocabulary, rephrase)
6. Commit — dgent hooks will clean trailers and emoji automatically