Writing and Content

Prompt-first procedure: Describe the outcome you want in your agent conversation. The agent should select and load the appropriate AIWG assets, explain material changes, request any needed approval, and report verification evidence. Exact commands and flags appear only in the CLI reference.

First time using AIWG? Begin with Install, Connect, and Verify. This guide assumes

AIWG is connected to the target project and your provider session can read the deployed context.

You want the AI to write the way you write — not in that recognizable AI voice that sounds the same as everything else.

AIWG's voice framework lets you define or select a writing style, then apply it consistently across everything the AI generates: documentation, blog posts, proposals, READMEs, emails, technical explainers, whatever.

Quickest path

Open your AI tool in the connected project and ask for one concrete writing result:

Write a README for this project in a clear, direct, technical tone

Rewrite this section so it sounds less like an AI wrote it

Write a blog post about what we built. Keep it conversational and specific — no generic hype.

Success means the draft preserves the facts, names the voice or guidance used, and gives you a short review note for anything that needs human judgment.

If you intentionally run a targeted deployment instead of the complete setup path, the agent-owned use operation remains the manual command for the writing and voice tools.

Built-in voice profiles

AIWG ships four ready-to-use voices:

Profile	When to use it
`technical-authority`	Documentation, API references, architecture explainers
`friendly-explainer`	Tutorials, onboarding, "how this works" guides
`executive-brief`	Stakeholder updates, proposals, board-ready summaries
`casual-conversational`	Blog posts, team updates, social content

Apply one in a session:

Use the technical-authority voice for everything in this session

Or specify it per request:

Rewrite this in the executive-brief style

Creating your own voice profile

If the built-in profiles don't match your brand or personal style, create a custom one.

Option 1 — From a sample of your existing writing:

/voice-create

Paste or link 3–5 samples of writing you like. AIWG analyzes them and generates a voice profile that captures the patterns.

Option 2 — Describe it:

Create a voice profile for our company blog: direct, slightly opinionated, no jargon,
short sentences, occasional dry humor. Technical audience but not academics.

The profile gets saved in `agentic/code/addons/voice-framework/voices/` and is available in future sessions.

Checking your content

After writing, run the validation:

/writing-validator path/to/content.md

This checks phrase patterns, passive voice, hedging and structure for editorial review. A match can be appropriate in context, including necessary uncertainty, quotations and technical terms.

The numeric `score`, `authenticityScore` and `aiPatternScore` fields are deprecated legacy heuristics, not evidence or probabilities of human authorship. The CLI retains threshold options and exit codes for compatibility; those are legacy diagnostics, not publication gates. Review the findings and retain useful wording rather than aim for zero highlights.

Rewriting existing content

If you have content that already sounds too "AI":

Rewrite this to sound more human. Keep all the information, change the delivery.

This reads like a template. Make it specific and concrete.

Strip the filler words and get to the point.

Or run the voice-apply skill across a whole file:

/voice-apply path/to/document.md the profile option technical-authority

Common writing tasks

Documentation:

Write API documentation for this function in technical-authority voice

Create a troubleshooting guide for the most common errors in this system

Blog and thought leadership:

Write a blog post about why we made this technical decision, written as a practitioner
talking to other practitioners

Proposals and briefs:

Summarize this project for an executive audience in two paragraphs

Write a proposal for adding authentication to this service — explain the problem,
options we considered, and what we recommend

Team communication:

Write a release announcement for this feature — internal team, casual tone

Working across sessions

Voice profiles persist — you don't need to re-specify them. Once you've run `/voice-create` or selected a profile, it's available by name in any session:

Use the "company-blog" voice for this

See the Voice Framework guide for advanced usage.