Configuration Reference | TechWrit AI Docs

TechWrit AI exports and imports configuration as a single JSON file (techwrit-config.json). This page documents every field so you can hand-edit a config, build one from scratch, or generate one programmatically.

Top-level structure

{
  "rules": [],
  "termSubs": [],
  "glossary": [],
  "customInst": "",
  "modeInstructions": {},
  "templates": []
}

All fields are optional on import — missing fields are left unchanged.

rules

An array of style rule objects. Each rule tells the AI to check for (and fix) a specific writing pattern.

{
  "id": "no-please",
  "label": "Avoid \"Please\"",
  "description": "Don't start instructions with \"Please\" — use direct, imperative language.",
  "active": true,
  "custom": false,
  "category": "Tone"
}

Field	Type	Required	Description
`id`	string	Yes	Unique identifier. Built-in IDs are fixed; custom rules should use a slug (e.g., `"no-exclamation-points"`).
`label`	string	Yes	Short display name shown in the Settings toggle list.
`description`	string	Yes	The instruction sent to the AI. Be specific — this is the actual rule text the model reads.
`active`	boolean	Yes	Whether the rule is included in the prompt. Set to `false` to keep a rule defined but disabled.
`custom`	boolean	Yes	`true` for user-created rules, `false` for built-in defaults.
`category`	string	No	Grouping label. Built-in rules use `"Tone"`, `"Clarity"`, `"Formatting"`, or `"Structure"`. Custom rules default to `"Custom"`.

Built-in rule IDs

no-please, no-and-or, title-case-titles, sentence-case-headings, no-may, ensure-vs-confirm, active-voice, present-tense, second-person, no-future-tense, parallel-structure, no-jargon-undefined, concise, no-there-is, consistent-terminology, no-redundancy

Example: adding a custom rule

{
  "id": "oxford-comma",
  "label": "Use Oxford commas",
  "description": "Always use a comma before the conjunction in a list of three or more items.",
  "active": true,
  "custom": true,
  "category": "Formatting"
}

termSubs

An array of terminology substitution objects. Each entry defines a preferred term and a list of terms to avoid.

{
  "preferred": "select",
  "avoid": ["click on", "click"]
}

Field	Type	Required	Description
`preferred`	string	Yes	The term to use instead.
`avoid`	string[]	Yes	One or more terms to flag and replace. Case-insensitive matching.

Default substitutions

[
  { "preferred": "select", "avoid": ["click on", "click"] },
  { "preferred": "enter", "avoid": ["type in", "input"] },
  { "preferred": "repository", "avoid": ["repo"] }
]

Example: adding substitutions

[
  { "preferred": "endpoint", "avoid": ["API route", "route"] },
  { "preferred": "run", "avoid": ["execute", "invoke"] },
  { "preferred": "set up", "avoid": ["setup", "configure"] }
]

glossary

An array of glossary entry objects. These give the AI semantic understanding of your product terms — not just spelling, but meaning.

{
  "term": "webhook",
  "definition": "An HTTP callback that sends real-time notifications to your application when an event occurs.",
  "synonyms": ["callback URL", "HTTP callback"]
}

Field	Type	Required	Description
`term`	string	Yes	The canonical term.
`definition`	string	Yes	A 1-2 sentence explanation. The AI uses this definition when writing about or reviewing the term.
`synonyms`	string[]	No	Alternative names. The AI flags these and suggests the canonical term instead.

Example: a product glossary

[
  {
    "term": "workspace",
    "definition": "The main editing environment where users compose input, select modes, and view AI output.",
    "synonyms": ["editor", "dashboard"]
  },
  {
    "term": "style rule",
    "definition": "A writing guideline that TechWrit AI enforces during review, rewrite, and generation.",
    "synonyms": ["writing rule", "lint rule"]
  },
  {
    "term": "context bar",
    "definition": "The horizontal bar above the input area containing Audience, Doc Type, Mode, and Prompt Library selectors."
  }
]

customInst

A string containing freeform instructions appended to every AI request. Use this for guidance that doesn't fit a single style rule.

{
  "customInst": "Our audience is enterprise IT admins. Keep paragraphs under 4 sentences. Always use Oxford commas."
}

Set to an empty string ("") to clear.

modeInstructions

An object mapping mode IDs to mode-specific instruction strings. When a mode has an entry here, it replaces the global customInst for that mode. Modes without an entry fall back to customInst.

{
  "modeInstructions": {
    "review": "Focus on security concerns and OWASP compliance.",
    "rewrite": "Always use Oxford commas. Prefer bullet lists for non-sequential content.",
    "simplify": "Target a 6th-grade reading level for this project."
  }
}

Key	Type	Description
Mode ID (e.g., `"review"`)	string	The instruction to use instead of `customInst` when this mode is active. See the Valid mode IDs table for all accepted keys.

Set a mode's value to an empty string or omit the key entirely to use the global customInst fallback.

This field is optional. If omitted or set to {}, the global customInst applies to all modes.

templates

An array of Prompt Library entry objects. Prompts pre-fill the input area with structured content containing placeholders.

{
  "id": "changelog-entry",
  "name": "Changelog Entry",
  "content": "Component: [name]\nDate: [date]\n\nChanges:\n- [change 1]\n- [change 2]\n\n=== OUTPUT FORMAT ===\nFormat as a changelog entry with a summary sentence followed by a bulleted list of changes.",
  "mode": "generate",
  "docType": "Release notes",
  "builtin": false
}

Field	Type	Required	Description
`id`	string	Yes	Unique identifier. Use a slug format (e.g., `"changelog-entry"`).
`name`	string	Yes	Display name shown in the Prompt Library dropdown.
`content`	string	Yes	The prompt body. Use `[placeholder]` syntax for fields the user fills in. Include an `=== OUTPUT FORMAT ===` section to guide the AI's response structure.
`mode`	string	No	Auto-set mode when the prompt is applied. See valid mode IDs below.
`docType`	string	No	Auto-set doc type when the prompt is applied. See valid doc types below.
`builtin`	boolean	Yes	Set to `false` for custom prompts. Built-in prompts (`true`) are managed by the app and cannot be overridden via import.

Prompt content tips

Use [bracketed placeholders] for fields the user fills in — e.g., [product name], [version number].
Add an === OUTPUT FORMAT === section at the end to instruct the AI on structure, headings, and format.
Use \n for line breaks in JSON (or actual newlines if editing in the Settings UI).

Valid mode IDs

Use these values for the mode field in Prompt Library entries:

ID	Label
`generate`	Write
`rewrite`	Rewrite
`review`	Review
`style-check`	Style Check
`simplify`	Simplify
`keywords`	Keywords
`glossary-gen`	Glossary
`code-to-docs`	Code to Docs
`user-guide`	User Guide
`explain`	Explain
`summarize`	Summarize
`expand`	Expand
`translate`	Translate
`outline`	Outline

Valid doc types

Use these values for the docType field in Prompt Library entries:

API reference, How-to guide, Conceptual overview, Tutorial, Release notes, Troubleshooting guide, README, Runbook, Migration guide, Architecture decision record, FAQ

Valid rule categories

Use these values for the category field in style rules:

Tone, Clarity, Formatting, Structure

Custom rules can use any category string — unrecognized values appear as their own group in Settings.

Complete example

A minimal but functional configuration:

{
  "rules": [
    { "id": "no-please", "label": "Avoid \"Please\"", "description": "Don't start instructions with \"Please\" — use direct, imperative language.", "active": true, "custom": false, "category": "Tone" },
    { "id": "active-voice", "label": "Prefer active voice", "description": "Use active voice unless passive is clearer or the actor is unknown.", "active": true, "custom": false, "category": "Tone" },
    { "id": "no-weasel-words", "label": "No weasel words", "description": "Avoid vague qualifiers like \"simply\", \"just\", \"easily\", and \"obviously\" — they undermine the reader's experience when the task isn't easy.", "active": true, "custom": true, "category": "Clarity" }
  ],
  "termSubs": [
    { "preferred": "select", "avoid": ["click on", "click"] },
    { "preferred": "enter", "avoid": ["type in", "input"] }
  ],
  "glossary": [
    {
      "term": "webhook",
      "definition": "An HTTP callback that sends real-time notifications to your application when an event occurs.",
      "synonyms": ["callback URL"]
    }
  ],
  "customInst": "Target a 9th-grade reading level. Use Oxford commas.",
  "modeInstructions": {
    "review": "Focus on security concerns and OWASP compliance.",
    "simplify": "Target a 6th-grade reading level for this project."
  },
  "templates": [
    {
      "id": "sdk-quickstart",
      "name": "SDK Quickstart",
      "content": "SDK name: [name]\nLanguage: [language]\nInstall command: [install command]\nMain use case: [what developers build with it]\n\n=== OUTPUT FORMAT ===\nFormat as a quickstart guide with: Overview, Prerequisites, Installation, First API call (with code example), Next steps.",
      "mode": "generate",
      "docType": "Tutorial",
      "builtin": false
    }
  ]
}