LoopyLuci Skill Error Recovery
Trigger: Use when following a skill's steps and a command fails, an error occurs, or the expected output doesn't appear.
Error Classification
| Error type | Example | Recovery strategy |
|---|---|---|
| Command not found | docker-compose: command not found |
Check installation, use v2 syntax |
| Permission denied | Permission denied (publickey) |
Check credentials, SSH keys, tokens |
| API changed | flag provided but not defined |
Tool was updated — adjust command |
| Wrong context | Not in a git repository |
Check working directory |
| Missing dependency | ModuleNotFoundError |
Install missing package |
| Network error | Could not resolve host |
Check connectivity, retry |
| Version mismatch | requires >=3.8 but you have 3.7 |
Check version requirements |
Recovery Decision Tree
Command from skill failed
│
┌────▼──────────────┐
│ Known error? │
│ (listed in skill's │──YES──► Apply skill's pitfall fix
│ Pitfalls section) │
└────┬──────────────┘
│ NO
┌────▼──────────────┐
│ Is error in the │
│ command syntax, │──YES──► Adjust to current version
│ not the approach? │
└────┬──────────────┘
│ NO
┌────▼──────────────┐
│ Is approach │
│ fundamentally │──YES──► Switch to different approach
│ wrong? │
└────┬──────────────┘
│ NO
┌────▼──────────────┐
│ Unfamiliar error │──YES──► Use general debugging skill
└────┬──────────────┘
│
└──► Report: patch the skill with this fix
Common Error Patterns
Pattern 1: Deprecated CLI
SKILL SAYS: docker-compose up -d
ERROR: 'docker-compose' is not recognized
FIX: docker compose up -d
(v2 removed the hyphen)
ACTION: Patch the skill
Pattern 2: Missing Setup Step
SKILL SAYS: aws s3 sync ./dist s3://bucket
ERROR: Unable to locate credentials
FIX: Run `aws configure` first, or set env vars
MISSING: Skill didn't mention authentication
ACTION: Add setup prerequisites to the skill
Pattern 3: Wrong Directory/Context
SKILL SAYS: cargo test
ERROR: could not find `Cargo.toml`
FIX: cd to the correct project directory first
MISSING: Skill assumed you were in the right dir
ACTION: Add `cd` step or context assumption to skill
Recovery Resources
When the skill doesn't have the fix, try:
1. ERROR MESSAGE → search terms
- Copy the exact error and search skills
- `web_search` the error message
2. --help flag
- `command --help` for current syntax
- Compare with skill's syntax
3. Fallback skills
- Load systematic-debugging
- Load a sibling skill (different approach, same goal)
4. General knowledge
- Apply general debugging principles
- Adapt the skill's approach to the real error
When to Patch the Skill
PATCH immediately if:
- The error is caused by tool evolution (not environment)
- Multiple users would hit the same issue
- The fix is a simple command substitution
DON'T PATCH if:
- The error is environment-specific (Python 3.7 on this machine)
- The error is user-specific (missing permissions)
- You're unsure if the fix is correct
PATCH with:
skill_manage(
action="patch",
name="skill-name",
old_string="old-command",
new_string="new-command"
)
Pitfalls
- Environment vs skill error: Is the command wrong, or is your environment misconfigured? Check before patching
- Multiple errors: Fix one error at a time — fixing everything at once can mask the root cause
- Partial success: Some steps worked, some didn't — document what failed, not the whole skill
- False positives: The command ran without error but produced wrong output — harder to catch than crashes
Verification
After recovery:
- Does the corrected approach work?
- Should the skill be patched with the fix?
- Did I document the error for future reference?