Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add ntorga/agent-starter-kit --skill loop-recoverygit clone --depth 1 https://github.com/ntorga/agent-starter-kitWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/ntorga/agent-starter-kit/loop-recovery)<a href="https://agentmods.dev/skills/ntorga/agent-starter-kit/loop-recovery"><img src="https://agentmods.dev/badge/skills/ntorga/agent-starter-kit/loop-recovery/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/ntorga/agent-starter-kit/loop-recovery"><img src="https://agentmods.dev/badge/skills/ntorga/agent-starter-kit/loop-recovery.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00016 | $0.00561 |
| Opus 5 | $0.00008 | $0.00280 |
| Sonnet 5 | $0.00003 | $0.00112 |
| Haiku 4.5 | $0.00002 | $0.00056 |
Grade A, and why
loop-recovery scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 38 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Purpose
When an agent has repeated the same failing action three times, it needs a concrete recovery procedure — not just the instruction to "reassess." This skill defines the recovery steps, the escalation path when the second approach also fails, and the detection patterns that catch loops before they waste effort.
Procedure
-
Stop and report. State what was attempted and what failed. Do not retry.
-
Check for oscillation. Before diagnosing, review the last several actions. If the agent has been alternating between two states — modifying file A, then file B, then file A again, or toggling between two error messages — the loop is structural, not incidental. Skip straight to step 5 on second detection.
-
Check for drift. If recent actions touched files unrelated to the stated goal, the agent is flailing. Acknowledge the drift, discard the unrelated changes, and refocus on the original objective before diagnosing.
-
Diagnose. Identify why the approach is not working. The answer must be a root cause, not a symptom — "the API rejects the payload because field X is required" is a diagnosis; "it returns 400" is a symptom.
-
Pivot. Propose a fundamentally different approach. Trivial variations (different whitespace, reordered arguments, minor rewording) do not count as a different approach. If no alternative exists, skip to step 7.
-
Execute the pivot. Attempt the new approach. If it succeeds, resume normal work. If it fails twice with the same result, proceed to step 7.
-
Abandon and hand off. Return a handoff explaining:
- The original approach and why it failed.
- The alternative approach and why it also failed (or why no alternative existed).
- What is needed to unblock (different tooling, human intervention, missing information).
Guardrails
- Never attempt a third approach. Two failed approaches show the problem needs human judgment or context the agent lacks.
- Never retry with debug commands, additional logging, or exploratory reads unless the new information would change the approach — not merely confirm the same failure.
- Never continue after detecting oscillation twice. An A→B→A→B cycle means the two approaches are undoing each other — a third attempt will repeat the pattern.
- Never treat drift (touching unrelated files) as progress. If the last action modified files outside the task scope, revert it before proceeding.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 38 lines · 16 tokens per session scan A 5e227df743da
loop-recovery is a skill published in the GitHub repository ntorga/agent-starter-kit (142 stars, last pushed 2d ago), licensed MIT. It adds 16 tokens to every session and 561 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-13.
Other skills, from other repositories
agent-framework-py-release
Use when cutting a Python release for the microsoft/agent-framework monorepo. Triggers on "bump py versions", "cut a python release", "prepare release PR for python", "release py packages", "bump python to X.Y.Z", or similar requests to bump Python package versions and prepare a release PR. Handles all four lifecycle…
python-package-management
Guide for managing packages in the Agent Framework Python monorepo, including creating new connector packages, versioning, and the lazy-loading pattern. Use this when adding, modifying, or releasing packages.
foundry-hosted-agent-validation
Step-by-step process for validating a Python Foundry hosted agent sample (under python/samples/04-hosting/foundry-hosted-agents/) end to end — running it locally (native runtime and azd ai agent run) and after deploying it to an Azure AI Foundry project with azd. Use this when asked to validate a hosted agent sample.
verify-samples-tool
How to use the verify-samples tool to run, verify, and manage sample definitions in the Agent Framework repository. Use this when adding, updating, or running sample verification.
build-and-test
How to build and test .NET projects in the Agent Framework repository. Use this when verifying or testing changes.
python-feature-lifecycle
Guidance for package and feature lifecycle in the Agent Framework Python codebase, including stage meanings, feature-stage decorators, feature enums, and how to move APIs from one stage to the next.