# AGENTS.md — Build guide for ProWritingAid

## Project scope
Open local prose, run transparent checks for repetition and sentence structure and review suggested rewrites as diffs. Keep original text local unless the user explicitly sends a selected passage to a model.

Catalogue verdict: yes. The core loop is small enough for a capable coding agent to produce a useful local version in one sitting. For ProWritingAid, run deep style reports, manuscript checks, and guided rewrites for long-form prose. The hard boundary is its mature report library, integrations, and fiction-specific analysis, plus polish, integrations, and proprietary language models.
Use the implementation prompt below to define the deliverable. Complete each phase's acceptance checks before extending the scope.

## Working agreement
- Inspect the repository and its existing instructions before choosing paths, dependencies or commands. Keep one coherent stack and explain changes to the proposed architecture.
- Plan a vertical slice that accepts a real input and produces the useful output described below. Persist only the state the prompt calls for; respect memory-only and upstream-managed workflows. Use fixtures only when they are clearly labelled.
- After scaffolding, document the actual install, development, check and build commands in README and keep them synchronized with the package or project manifest. Do not report commands as successful unless they ran.
- Work in small steps. At handoff, list implemented flows, checks actually performed, remaining blockers, and any credentials or provider setup the owner must supply.
- Do not publish, spend money, contact customers, delete source data or run irreversible migrations without the project owner's authorization.

## Prerequisites
- Node and the Rust toolchain, a supported desktop build environment and an explicitly selected data folder. Remote sync is outside the initial scope.
- Implementation components: Tauri, React and TypeScript for the desktop interface with a narrowly scoped Rust filesystem bridge. SQLite for structured state and search; user-owned files for original documents and attachments.
- Scope boundary: its mature report library, integrations, and fiction-specific analysis; cross-device sync

## Stack and architecture
- Tauri, React and TypeScript for the desktop interface with a narrowly scoped Rust filesystem bridge.
- SQLite for structured state and search; user-owned files for original documents and attachments.
- Domain model: manuscripts, sentence spans, deterministic style findings, rewrite suggestions and accepted edits

## Security and data integrity
- Limit Tauri capabilities to selected folders and commands. Render imported content without executable HTML, and never expose arbitrary shell execution to the webview.
- Correctness boundary: Style rules describe choices rather than universal correctness; accepted edits remain undoable and unsupported grammar claims are labeled.
- Save edits atomically with stable IDs and revision history. Detect external file changes and offer a conflict copy instead of last-write-wins data loss.
- Provide a portable folder export with a JSON manifest and original files. Restore to a separate folder; rebuild disposable indexes from sources.

## Agent implementation rules
- Project rule — data model: manuscripts, sentence spans, deterministic style findings, rewrite suggestions and accepted edits
- Project rule — preserve this invariant: Style rules describe choices rather than universal correctness; accepted edits remain undoable and unsupported grammar claims are labeled.
- Project rule — acceptance evidence: Reject a rewrite without changing the manuscript; edit an earlier paragraph and invalidate stale finding offsets instead of highlighting the wrong words.

## Optional agent skills and references
- Optional external skill: [vercel-react-best-practices](https://github.com/vercel-labs/agent-skills/blob/main/skills/react-best-practices/SKILL.md) — Improve React and Next.js data fetching, rendering, bundle size and server performance. Review its instructions and compatibility before use; it does not grant deployment, data-access or publication permission.
- Optional external skill: [web-design-guidelines](https://github.com/vercel-labs/agent-skills/blob/main/skills/web-design-guidelines/SKILL.md) — Review web interfaces for accessibility, keyboard focus, forms, navigation and interaction quality. Review its instructions and compatibility before use; it does not grant deployment, data-access or publication permission.
- Optional external skill: [sharp-edges](https://github.com/trailofbits/skills/blob/main/plugins/sharp-edges/skills/sharp-edges/SKILL.md) — Review security-sensitive APIs and configuration for dangerous defaults and easy-to-misuse interfaces. Review its instructions and compatibility before use; it does not grant deployment, data-access or publication permission.

Read the linked SKILL.md and its dependencies before adding a skill. Select only the skills matching this project's runtime and task; their documentation does not supply API access, credentials or approval to perform external actions. Pin the reviewed revision where the tool supports it. Follow the chosen agent's documented project-level installation mechanism.

## Distribution ideas
These are optional planning notes. Obtain the owner's approval before publishing or contacting anyone.
- Demonstrate the actual ProWritingAid-inspired workflow with owned or clearly labeled sample data: Open local prose, run transparent checks for repetition and sentence structure and review suggested rewrites as diffs. Keep original text local unless the user explicitly sends a selected passage to a model.
- Publish a reproducible walkthrough with this observable result: Reject a rewrite without changing the manuscript; edit an earlier paragraph and invalidate stale finding offsets instead of highlighting the wrong words.
- Explain who can operate this scoped tool, its setup and ongoing costs, and these remaining product gaps: its mature report library, integrations, and fiction-specific analysis; cross-device sync Avoid guaranteed savings, performance scores or implied endorsement.

## Engineering roadmap
1. Phase 1 — Scope and fixtures. Implement this bounded workflow: Open local prose, run transparent checks for repetition and sentence structure and review suggested rewrites as diffs. Keep original text local unless the user explicitly sends a selected passage to a model. Record prerequisites, select representative user-owned fixtures and document the unsupported features: its mature report library, integrations, and fiction-specific analysis; cross-device sync
2. Phase 2 — Durable model. Model manuscripts, sentence spans, deterministic style findings, rewrite suggestions and accepted edits Add migrations or a versioned document format, explicit validation, stable IDs and a visible import-error report. Preserve this rule: Style rules describe choices rather than universal correctness; accepted edits remain undoable and unsupported grammar claims are labeled.
3. Phase 3 — Complete the first useful path. Implement the workflow's input, review and output interface, with clear controls and explicit empty/error states. Save edits atomically with stable IDs and revision history. Detect external file changes and offer a conflict copy instead of last-write-wins data loss.
4. Phase 4 — Permissions and integration failure. Limit Tauri capabilities to selected folders and commands. Render imported content without executable HTML, and never expose arbitrary shell execution to the webview. Request integration credentials and permissions only for the enabled feature; show a disconnected state instead of mock results.
5. Phase 5 — Portable handoff. Provide a portable folder export with a JSON manifest and original files. Restore to a separate folder; rebuild disposable indexes from sources. Include setup, operating limits, fixture walkthrough and shutdown/restart instructions in the README.
6. Phase 6 — Acceptance scenarios. Reject a rewrite without changing the manuscript; edit an earlier paragraph and invalidate stale finding offsets instead of highlighting the wrong words. Repeat the workflow after restart and with a denied permission or unavailable dependency; show recoverable failure rather than a success placeholder.

## Paid-product capabilities outside this build
- its mature report library, integrations, and fiction-specific analysis
- cross-device sync
- large proprietary style models
- browser and office-suite extensions
- team style guides

## Implementation prompt
WORKING SLICE
Open local prose, run transparent checks for repetition and sentence structure and review suggested rewrites as diffs. Keep original text local unless the user explicitly sends a selected passage to a model.

Build this scoped ProWritingAid-inspired workflow with a documented data model and visible failure states.

Architecture
- Tauri, React and TypeScript for the desktop interface with a narrowly scoped Rust filesystem bridge.
- SQLite for structured state and search; user-owned files for original documents and attachments.

Prerequisites and limits
Node and the Rust toolchain, a supported desktop build environment and an explicitly selected data folder. Remote sync is outside the initial scope.
Outside this release: its mature report library, integrations, and fiction-specific analysis; cross-device sync

Data model and correctness
manuscripts, sentence spans, deterministic style findings, rewrite suggestions and accepted edits
Invariant: Style rules describe choices rather than universal correctness; accepted edits remain undoable and unsupported grammar claims are labeled.
Save edits atomically with stable IDs and revision history. Detect external file changes and offer a conflict copy instead of last-write-wins data loss.

Security and privacy
Limit Tauri capabilities to selected folders and commands. Render imported content without executable HTML, and never expose arbitrary shell execution to the webview.

Recovery and export
Provide a portable folder export with a JSON manifest and original files. Restore to a separate folder; rebuild disposable indexes from sources.

Implementation order
1. Phase 1 — Scope and fixtures. Implement this bounded workflow: Open local prose, run transparent checks for repetition and sentence structure and review suggested rewrites as diffs. Keep original text local unless the user explicitly sends a selected passage to a model. Record prerequisites, select representative user-owned fixtures and document the unsupported features: its mature report library, integrations, and fiction-specific analysis; cross-device sync
2. Phase 2 — Durable model. Model manuscripts, sentence spans, deterministic style findings, rewrite suggestions and accepted edits Add migrations or a versioned document format, explicit validation, stable IDs and a visible import-error report. Preserve this rule: Style rules describe choices rather than universal correctness; accepted edits remain undoable and unsupported grammar claims are labeled.
3. Phase 3 — Complete the first useful path. Implement the workflow's input, review and output interface, with clear controls and explicit empty/error states. Save edits atomically with stable IDs and revision history. Detect external file changes and offer a conflict copy instead of last-write-wins data loss.
4. Phase 4 — Permissions and integration failure. Limit Tauri capabilities to selected folders and commands. Render imported content without executable HTML, and never expose arbitrary shell execution to the webview. Request integration credentials and permissions only for the enabled feature; show a disconnected state instead of mock results.
5. Phase 5 — Portable handoff. Provide a portable folder export with a JSON manifest and original files. Restore to a separate folder; rebuild disposable indexes from sources. Include setup, operating limits, fixture walkthrough and shutdown/restart instructions in the README.
6. Phase 6 — Acceptance scenarios. Reject a rewrite without changing the manuscript; edit an earlier paragraph and invalidate stale finding offsets instead of highlighting the wrong words. Repeat the workflow after restart and with a denied permission or unavailable dependency; show recoverable failure rather than a success placeholder.

Acceptance
Reject a rewrite without changing the manuscript; edit an earlier paragraph and invalidate stale finding offsets instead of highlighting the wrong words.
Use real source data or clearly labeled fixtures. Explain unsupported input and provider failures; do not fabricate analytics, delivery receipts, accuracy claims or security guarantees.

Optional agent guidance
Optional external skill: [vercel-react-best-practices](https://github.com/vercel-labs/agent-skills/blob/main/skills/react-best-practices/SKILL.md) — Improve React and Next.js data fetching, rendering, bundle size and server performance. Review its instructions and compatibility before use; it does not grant deployment, data-access or publication permission.
Optional external skill: [web-design-guidelines](https://github.com/vercel-labs/agent-skills/blob/main/skills/web-design-guidelines/SKILL.md) — Review web interfaces for accessibility, keyboard focus, forms, navigation and interaction quality. Review its instructions and compatibility before use; it does not grant deployment, data-access or publication permission.
Optional external skill: [sharp-edges](https://github.com/trailofbits/skills/blob/main/plugins/sharp-edges/skills/sharp-edges/SKILL.md) — Review security-sensitive APIs and configuration for dangerous defaults and easy-to-misuse interfaces. Review its instructions and compatibility before use; it does not grant deployment, data-access or publication permission.
Project rule — data model: manuscripts, sentence spans, deterministic style findings, rewrite suggestions and accepted edits
Project rule — preserve this invariant: Style rules describe choices rather than universal correctness; accepted edits remain undoable and unsupported grammar claims are labeled.
Project rule — acceptance evidence: Reject a rewrite without changing the manuscript; edit an earlier paragraph and invalidate stale finding offsets instead of highlighting the wrong words.

## Completion evidence
Demonstrate the prompt's acceptance scenarios against the scoped workflow. Include setup from a clean checkout and failure recovery. Check persistence across restart and export/restore only for the state the prompt says to store; for memory-only tools, confirm that temporary content is discarded as specified. Record actual results and remaining limitations. A detailed plan alone does not establish a working replacement.