Add AI chat history, conversation search, and per-user personalization by Copilot · Pull Request #16 · QuickMash/Komli

Copilot · 2026-03-11T21:48:00Z

The AI had no memory of prior messages — every request was sent to Ollama as a single-turn exchange with no conversation context. Additionally, there was no way to search past conversations.

Changes

`processing/ai.py` — Context-aware `send()`

Accepts history: list (prior DB messages) and user_name: str
Inserts history turns between the system prompt and the current user message before calling Ollama
Personalizes system prompt with user's name when available

# Before: always a single-turn exchange
messages = [system_message, user_message]

# After: full conversation context
messages = [system_message, *history_turns, user_message]

`app.py` — Pre-fetch history before AI call; search endpoint

respond() now resolves the active conversation and fetches its messages before calling ai.send(), then passes them as history
New GET /api/search?q=<query> endpoint scoped to the authenticated user

`login/server.py` — `search_conversations()`

LIKE search across conversation titles and message content
User-supplied query is escaped for %, _, and \ before pattern construction to prevent wildcard manipulation

`templates/index.html` — Sidebar search UI

Search input added at top of sidebar (logged-in users only)
300 ms debounced fetch to /api/search, results rendered as clickable conversation items

Original prompt

This section details on the original issue you should resolve

<issue_title>No history for ai</issue_title>
<issue_description>The ai cannot see history of messages</issue_description>

<agent_instructions>The ai assistant doesnt have a chat history, and it makes it painful to use, additionally please add search, and other per user stuff.</agent_instructions>

Comments on the Issue (you are @copilot in this section)

Fixes No history for ai #14

🔒 GitHub Advanced Security automatically protects Copilot coding agent pull requests. You can protect all pull requests by enabling Advanced Security for your repositories. Learn more about Advanced Security.

github-actions

Welcome to Komli!

Co-authored-by: QuickMash <106212829+QuickMash@users.noreply.github.com>

Copilot

Pull request overview

Adds conversation memory and user-scoped conversation search to Komli so AI responses can incorporate prior turns and users can find past chats.

Changes:

Passes DB conversation history + user name into processing.ai.send() to build a context-aware Ollama prompt.
Adds /api/search endpoint and SQLite query helper to search titles/message content for the authenticated user.
Adds a sidebar search UI with debounced requests and clickable results.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 4 comments.

File	Description
`processing/ai.py`	Expands `send()` to include prior turns and personalize the system prompt.
`app.py`	Fetches conversation history before AI calls; adds `/api/search` API endpoint.
`login/server.py`	Implements `search_conversations()` with LIKE escaping and result shaping.
`templates/index.html`	Adds sidebar search input, styles, and debounced client-side search rendering.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-03-12T02:01:58Z

+    messages = [system_message]
+
+    # Include prior conversation turns so the AI has memory of the chat
+    if history:
+        for msg in history:
+            role = 'user' if msg.get('message_type') == 'user' else 'assistant'
+            messages.append({'role': role, 'content': msg['content']})
+
+    messages.append({'role': 'user', 'content': user_input})


send() now appends the entire conversation history to the Ollama prompt. Since get_conversation_messages() returns all messages in a conversation, prompts can grow without bound and eventually exceed the model context window or cause very slow requests. Consider truncating/summarizing history (e.g., last N turns or last N characters/tokens) before calling Ollama, ideally using a configurable limit.

Copilot · 2026-03-12T02:01:58Z

+    if user_email:
+        try:
+            if conversation_id:
+                conversation_record = server.get_conversation(int(conversation_id), user_email)
+                if conversation_record:
+                    target_conversation_id = int(conversation_id)
+                else:
+                    target_conversation_id = server.get_or_create_active_conversation(user_email)
+            else:
+                target_conversation_id = server.get_or_create_active_conversation(user_email)
+
+            # Fetch existing messages so the AI sees the full conversation history
+            if target_conversation_id:
+                history = server.get_conversation_messages(target_conversation_id)
+        except Exception:
+            pass


The history-fetch/conversation-resolution block swallows all exceptions and then proceeds, which can silently disable both history and DB logging for logged-in users (e.g., non-integer conversation_id, DB errors). Since this code is critical to the new “AI has memory” behavior, consider handling expected errors (like ValueError) with a fallback to the active conversation and logging unexpected exceptions so failures are visible.

Copilot · 2026-03-12T02:01:58Z

+            resultsSection.style.display = 'block';
+            resultsList.innerHTML = '<div class="text-center py-2"><i class="fas fa-spinner fa-spin"></i></div>';
+
+            fetch('/api/search?q=' + encodeURIComponent(query))


The search UI calls fetch('/api/search?...') using an absolute root path. The backend routes are prefixed by web_dir (configurable) and the app is explicitly designed to run under a non-root base path; this call (and any new calls) will break when web_dir isn’t /. Consider using a base path value rendered from the template (or switching to a relative URL) so the request is correctly scoped under web_dir.

Suggested change

fetch('/api/search?q=' + encodeURIComponent(query))

fetch('api/search?q=' + encodeURIComponent(query))

Copilot · 2026-03-12T02:01:59Z

+def send(user_input: str, history: list = None, user_name: str = None) -> str:
+    name_part = f" You are talking to {user_name}." if user_name else ""
    system_message = {
        'role': 'system',
-        'content': f'You are {name}. {sys_prompt} Version: {version}. You are only allowed to speak with markdown formatting. Begin normal messages with ` and end them with `'
+        'content': f'You are {name}. {sys_prompt}{name_part} Version: {version}. You are only allowed to speak with markdown formatting. Begin normal messages with ` and end them with `'
    }


user_name comes from user-controlled profile data and is interpolated directly into the system prompt. This enables prompt-injection at the highest priority (system) level (e.g., a display name containing instructions/newlines/backticks), which can undermine safety constraints and may increase the risk of the model emitting unsafe HTML that is later marked safe via Markup(). Consider sanitizing/normalizing the name (length limit, strip newlines/control chars, quote/escape) and/or moving the personalization into a lower-priority message instead of the system prompt.

Initial plan

e491e27

Copilot AI assigned Copilot and QuickMash Mar 11, 2026

Copilot started work on behalf of QuickMash March 11, 2026 21:48 View session

github-actions Bot reviewed Mar 11, 2026

View reviewed changes

Add AI chat history, conversation search, and per-user personalization

656fb9b

Co-authored-by: QuickMash <106212829+QuickMash@users.noreply.github.com>

Copilot AI changed the title ~~[WIP] Add chat history feature for AI assistant~~ Add AI chat history, conversation search, and per-user personalization Mar 11, 2026

Copilot finished work on behalf of QuickMash March 11, 2026 21:54

QuickMash marked this pull request as ready for review March 12, 2026 01:56

Copilot AI review requested due to automatic review settings March 12, 2026 01:57

QuickMash merged commit 80a513d into main Mar 12, 2026
5 checks passed

Copilot started reviewing on behalf of QuickMash March 12, 2026 01:57 View session

Copilot AI reviewed Mar 12, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add AI chat history, conversation search, and per-user personalization#16

Add AI chat history, conversation search, and per-user personalization#16
QuickMash merged 2 commits into
mainfrom
copilot/add-chat-history-feature

Copilot AI commented Mar 11, 2026 •

edited

Loading

Uh oh!

github-actions Bot left a comment

Uh oh!

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Mar 12, 2026

Uh oh!

Copilot AI Mar 12, 2026

Uh oh!

Copilot AI Mar 12, 2026

Uh oh!

Copilot AI Mar 12, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

	fetch('/api/search?q=' + encodeURIComponent(query))
	fetch('api/search?q=' + encodeURIComponent(query))

Conversation

Copilot AI commented Mar 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Changes

processing/ai.py — Context-aware send()

app.py — Pre-fetch history before AI call; search endpoint

login/server.py — search_conversations()

templates/index.html — Sidebar search UI

Comments on the Issue (you are @copilot in this section)

Uh oh!

github-actions Bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Mar 12, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 12, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 12, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 12, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Copilot AI commented Mar 11, 2026 •

edited

Loading

`processing/ai.py` — Context-aware `send()`

`app.py` — Pre-fetch history before AI call; search endpoint

`login/server.py` — `search_conversations()`

`templates/index.html` — Sidebar search UI