- Home
- Skills
- Agents & Orchestration
- agent-debug-loop
Agent Debug Loop
Structured self-debugging for AI agents that are stuck, looping, or burning tokens without progress.
$7.99
Works with the AI Agents you already use
agent-debug-loop
Example session with this skill installed
My agent has called the same edit tool three times with a stale hash and is now looping without progress. Help it recover.
- Read your context and instructions
- Compiled the agent-debug-loop
Failure Capture
- Goal: apply patch to foo.ts
- Error: hash mismatch on edit
- Repeated pattern: 3 identical PUTs
- Assumptions to verify: file changed since read
Root-Cause Diagnosis
Pattern: stale anchor. Evidence: fresh read shows line 42 changed by a concurrent writer.
Contained Recovery
Re-read foo.ts, re-issue the edit with the new hash, verify with one read.
Introspection Report
- Failure class: stale anchor
- Recovery applied: fresh read + re-anchored edit
- Outcome: resolved
- Prevention: always re-read after a hash mismatch before retrying.
Connects securely to your tools. The creator never sees your data.
About this skill
When an agent run fails repeatedly, loops on the same tools, consumes tokens without progress, or drifts off task, this skill replaces blind retries with a four-phase workflow: capture failure state to a persistent artifact, diagnose against seven known failure patterns (stale anchors, path confusion, dependency gaps, context bloat, tool misuse, state races, loop locks), apply the smallest contained recovery action, and produce an introspection report with a prevention rule. Includes an anti-rationalization table that names the four excuses agents use to skip diagnosis. Works in any agent harness: no scripts, no dependencies, no network access.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Agents & Orchestration.
Workflow guide only; it does not automatically reset a harness or enforce loop limits.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 5 days ago
Requires
Nothing to install.
Permissions
No special permissions declared or detected