ChatGPT

Build me a personal chat client on the OpenAI and Anthropic APIs to replace the
text-chat part of ChatGPT Plus. Requirements:

- A local web app: Node + Express + better-sqlite3, one page with a streaming
  reply pane (SSE), a model picker, and a conversation sidebar.
- OpenAI and Anthropic keys in .env; system prompt editable per conversation.
- Conversations stored in SQLite with FTS5 search; export any thread to
  Markdown.
- File upload for text and PDF (pdf-parse); contents go straight into the
  context window, no vector database.
- A usage footer: tokens and estimated cost per conversation and per month,
  computed from the usage field in API responses, so I can compare real spend
  against the $20 flat fee.
- Localhost only. No accounts, no telemetry, everything stays on my machine
  except the API calls.
- Out of scope: image generation, voice mode, browsing agents, memory, and
  mobile apps. Do not attempt any of them; that gap is the subscription.
- README: where to get each key, links to current per-token pricing, and state
  plainly that this replaces text chat only, the frontier model itself cannot
  be rebuilt.

# AGENTS.md — Build guide for ChatGPT

## Project scope
Build a chat interface that calls OpenAI/Anthropic/local models, stores conversations, and adds file upload/tools.

Catalogue verdict: no. You can build a chat UI over APIs, but you cannot solo-recreate the model, multimodal stack, tools, memory/product layer, safety systems, and scale.
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- Do not publish, spend money, contact customers, delete source data or run irreversible migrations without the project owner's authorization.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
- LLM API key or local model
- hosted app
- vector store if retrieval
- auth/storage
- tool integrations

## Stack and architecture
- A local web app: Node + Express + better-sqlite3, one page with a streaming reply pane (SSE), a model picker, and a conversation sidebar.
- Model original assets, edit operations, previews, and exported versions; store source IDs and timestamps for each.
- Interface for this general AI assistant workflow: a local web page with input, progress, review, and export views.

## Security and data integrity
- If LLM API key or local model is unavailable, keep the source record and show a recoverable error instead of a fabricated result.
- Keep API keys and OAuth secrets in .env; redact them from logs, generated files, and error responses.
- State the practical limit: frontier models.

## Agent implementation rules
- Project rule, data: Model original assets, edit operations, previews, and exported versions; store source IDs and timestamps for each.
- Project rule, behavior: A usage footer: tokens and estimated cost per conversation and per month, computed from the usage field in API responses, so I can compare real spend against the $20 flat fee.
- Project rule, recovery: If LLM API key or local model is unavailable, keep the source record and show a recoverable error instead of a fabricated result.

## Optional agent skills and references
No installable skill is listed for this guide; use the project rules and the stack documentation.

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 this general AI assistant operation with sample or user-owned data: A usage footer: tokens and estimated cost per conversation and per month, computed from the usage field in API responses, so I can compare real spend against the $20 flat fee.
- Show a concrete result or export: Conversations stored in SQLite with FTS5 search; export any thread to Markdown.
- State the practical limit: frontier models.

## Engineering roadmap
1. Phase 1, architecture and data: A local web app: Node + Express + better-sqlite3, one page with a streaming reply pane (SSE), a model picker, and a conversation sidebar. Model original assets, edit operations, previews, and exported versions; store source IDs and timestamps for each.
2. Phase 2, implement: A usage footer: tokens and estimated cost per conversation and per month, computed from the usage field in API responses, so I can compare real spend against the $20 flat fee.
3. Phase 3, implement: Build a chat interface that calls OpenAI/Anthropic/local models, stores conversations, and adds file upload/tools.
4. Phase 4, review and output: General AI assistant for writing, coding, research, images, and workflows. Conversations stored in SQLite with FTS5 search; export any thread to Markdown.
5. Phase 5, recovery and acceptance: If LLM API key or local model is unavailable, keep the source record and show a recoverable error instead of a fabricated result. Verify this invariant with a saved fixture: An invalid input or interrupted operation must retain the source and show a recoverable state; exported records must reload with the same IDs. State the practical limit: frontier models.

## Paid-product capabilities outside this build
- frontier models
- multimodal features
- tool ecosystem
- reliability
- safety
- product updates
- mobile apps

## Implementation prompt
Build me a personal chat client on the OpenAI and Anthropic APIs to replace the
text-chat part of ChatGPT Plus. Requirements:

- A local web app: Node + Express + better-sqlite3, one page with a streaming
  reply pane (SSE), a model picker, and a conversation sidebar.
- OpenAI and Anthropic keys in .env; system prompt editable per conversation.
- Conversations stored in SQLite with FTS5 search; export any thread to
  Markdown.
- File upload for text and PDF (pdf-parse); contents go straight into the
  context window, no vector database.
- A usage footer: tokens and estimated cost per conversation and per month,
  computed from the usage field in API responses, so I can compare real spend
  against the $20 flat fee.
- Localhost only. No accounts, no telemetry, everything stays on my machine
  except the API calls.
- Out of scope: image generation, voice mode, browsing agents, memory, and
  mobile apps. Do not attempt any of them; that gap is the subscription.
- README: where to get each key, links to current per-token pricing, and state
  plainly that this replaces text chat only, the frontier model itself cannot
  be rebuilt.

## 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.