memory-troubleshoot
Diagnose and resolve common agent-memory issues safely. This skill never runs verification commands automatically and always asks for confirmation before edits or restarts.
Works with
Agent Skills format with YAML frontmatter. Claude Code reads it as-is.
---
name: "memory-troubleshoot"
description: "Diagnose and resolve common agent-memory issues safely. This skill never runs verification commands automatically and always asks for confirmation before edits or restarts."
license: "MIT"
---
# Memory Troubleshoot Skill
Diagnose and resolve common agent-memory issues safely. This skill never runs
verification commands automatically and always asks for confirmation before
edits or restarts.
## When to Use
- Daemon will not start
- Commands are not found
- Connection refused
- No events captured
- Summarization failures
## Safety Rules
- Do not edit files or restart services without confirmation
- Provide verification commands only
- Keep fixes scoped to the reported issue
## Diagnostic Flow
### Step 1: Identify the symptom
```
Which issue are you seeing?
1. "command not found" for memory-daemon or memory-ingest
2. "connection refused" or cannot query daemon
3. Daemon starts then stops
4. No events captured
5. Summarization not working
6. Other / not sure
Enter selection [1-6]:
```
### Step 2: Provide targeted diagnostics (commands only)
**1) Command not found**
```
which memory-daemon
which memory-ingest
echo $PATH
```
**2) Connection refused**
```
memory-daemon status
lsof -i :50051
memory-daemon query --endpoint http://[::1]:50051 root
```
**3) Daemon stops immediately**
```
memory-daemon status
tail -50 ~/Library/Logs/memory-daemon/daemon.log
tail -50 ~/.local/state/memory-daemon/daemon.log
```
**4) No events captured**
```
grep -n "memory-ingest" ~/.claude/hooks.yaml
which memory-ingest
memory-daemon status
```
**5) Summarization failing**
```
echo "OPENAI: ${OPENAI_API_KEY:+set}"
echo "ANTHROPIC: ${ANTHROPIC_API_KEY:+set}"
```
### Step 3: Offer safe fixes (with confirmation)
For each fix, ask:
```
I can apply the following fix:
[describe the change]
Proceed? (yes/no)
```
Examples:
**PATH update**
```
Add ~/.local/bin to PATH in your shell profile
```
**Restart daemon**
```
Stop and restart the daemon:
memory-daemon stop
memory-daemon start
```
**Fix hooks file**
```
Add memory-ingest handler to ~/.claude/hooks.yaml
```
### Step 4: Provide verification commands (optional)
```
memory-daemon status
memory-daemon query --endpoint http://[::1]:50051 root
```
## Common Fix Playbooks
### "command not found"
- Ensure binaries are installed
- Add install directory to PATH
- If missing, re-run `memory-install`
### "connection refused"
- Confirm daemon is running
- Check port conflicts on 50051
- Update `grpc_port` in config if needed
### "no events captured"
- Confirm hooks file contains `memory-ingest`
- Confirm `memory-ingest` is on PATH
- Restart the agent session after hook changes
### "summarization not working"
- Set provider API key
- Ensure config provider matches key
## Escalation
If diagnostics do not resolve the issue, recommend:
- Re-run `memory-verify` for a full check
- Re-run `memory-install` with a fresh setupMore General & Other skills
find-skills
vercel-labs/skills
Helps users discover and install agent skills when they ask questions like "how do I do X", "find a skill for X", "is there a skill that can...", or express interest in extending capabilities. This skill should be used when the user is looking for functionality that might exist as an installable skill.
grill-me
mattpocock/skills
A relentless interview to sharpen a plan or design.
grill-with-docs
mattpocock/skills
A relentless interview to sharpen a plan or design, which also creates docs (ADR's and glossary) as we go.

