Feature Request: Graceful handling of trivial/edge-case conversations

[spm1001]

Feature Request: Graceful handling of trivial/edge-case conversations

Context

After heavy usage (3000+ archived conversations), I noticed sync reporting Summarized: 0 despite processing conversations. After digging in, I found several classes of files that parse to 0 exchanges and block the queue.

Files that could be handled more gracefully

1. Empty files (0 bytes): Sessions started then immediately closed. Common when testing or fat-fingering.

2. Trivial sessions (<10 lines): Quick commands, failed sessions, etc. Not worth LLM summarization.

3. Agent files (agent-*.jsonl): Subagent sessions from Task tool. Often just assistant messages without a user prompt.

4. Summarizer's own sessions: The CLI folder accumulates sessions from summary generation itself.

Current behavior

These files are selected for summarization, parser returns 0 exchanges, loop silently continues. User sees "Generating summaries for 10" then "Summarized: 0" with no indication why.

Possible improvements

1. Skip empty/trivial files earlier: Check file size before adding to candidates, auto-create stub summaries like "Empty session" or "Trivial conversation"

2. Better feedback: Log when skipping files so users understand what's happening

3. Filter known-problematic patterns: Skip the episodic-memory CLI's own archive folder

Workaround

I wrote a script to stub these files manually:

# Stub empty and trivial files
for f in ~/.config/superpowers/conversation-archive/*/*.jsonl; do
  summary="${f%.jsonl}-summary.txt"
  if [ ! -f "$summary" ]; then
    lines=$(wc -l < "$f" 2>/dev/null || echo 0)
    if [ "$lines" -lt 10 ]; then
      echo "Trivial session" > "$summary"
    fi
  fi
done

Happy to help with a PR if you point me in the right direction.

Environment

• episodic-memory: 1.0.15 (via plugin)
• Claude Code: 2.0.75
• macOS 15.x

[coderabbitai]

📝 CodeRabbit Plan Mode

Generate an implementation plan and prompts that you can use with your favorite coding agent.

• Create Plan

Examples

• Example 1
• Example 2

---

🔗 Similar Issues

Possible Duplicates

• Bug Report: Agent conversation files not being summarized #22

Related Issues

• Summary generation fails when running episodic-memory sync #5
• Episodic memory appears empty - sync needs to run automatically or prompt user on first use #27
• agent conversations are incorrectly indexed #16

🔗 Related PRs

obra/lace#230 - Prompt improvement [merged]
obra/lace#296 - Convert session creation to inline form with chat input pre-filling [closed]
#15 - Fix: Add exclusion marker to summarizer prompts [merged]
#37 - feat: Add API configuration support for summarization [merged]

👤 Suggested Assignees

• one1zero1one
• obra
• ericgjerde
• kenryu42
• Jehu

---

🧪 Issue enrichment is currently in open beta.

You can configure auto-planning by selecting labels in the issue_enrichment configuration.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord or schedule a call!

[techjoec]

I was working on the same too. Was trying to find an approach I liked..Ended up with thousands of junk sessions myself too. Sick as a dog today with a fever so lagging on it but any approach to avoid bloat without value is good!

 Add validation before copying:

 // In sync.ts, before copyIfNewer()
 function hasIndexableContent(filePath: string): boolean {
   const content = fs.readFileSync(filePath, 'utf-8');
   return content.includes('"type":"user"') && content.includes('"type":"assistant"');
 }

 // Only copy if indexable
 if (hasIndexableContent(srcFile)) {
   const wasCopied = copyIfNewer(srcFile, destFile);
   // ...
 }