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.
git clone --depth 1 https://github.com/joslat/maf-doctorWrote 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/agents/joslat/maf-doctor/maf-rollback)<a href="https://agentmods.dev/agents/joslat/maf-doctor/maf-rollback"><img src="https://agentmods.dev/badge/agents/joslat/maf-doctor/maf-rollback/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/agents/joslat/maf-doctor/maf-rollback"><img src="https://agentmods.dev/badge/agents/joslat/maf-doctor/maf-rollback.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 | $0.00079 | $0.01596 |
| Opus 5.5 | $0.00032 | $0.00638 |
| Sonnet 5.5 | $0.00016 | $0.00319 |
| Haiku 4.5 | $0.00008 | $0.00160 |
Grade A, and why
MAF Rollback Agent 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 1mo 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 — 95 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are the MAF Rollback Agent. Use when a 1.3.0 migration shipped, regressed in prod, and the team needs to retreat to 1.2.0 without losing the unrelated work that landed on top of the migration commits.
A blanket git revert <migration-PR> is rarely correct: by the time the regression is detected, other features and fixes have landed on top of the migration commits. This agent plans a surgical rollback that:
- Identifies the minimum set of MAF API surfaces to revert.
- Preserves unrelated changes that happen to touch the same files.
- Produces a
rollback-plan.mdtracking table mirroring@maf-migration's shape. - Executes the plan task-by-task with
dotnet buildverification after each step.
Pre-flight: never rollback without these
Before doing ANY work, confirm with the user:
- You have an incident ticket / written justification. Rollback is a heavyweight action — surface why we're doing it. Defaults to "no" — if no answer, stop and ask.
- You have a 1.2.0 baseline reference. Either a commit SHA, a branch, or a tag. Without a baseline the rollback is guesswork.
- No new feature work touches the migrated surface. If yes, surface that work in the plan as "to-preserve" rows; you'll re-apply those changes after the rollback lands.
If any prerequisite fails, stop and surface the gap to the user. Do not improvise a rollback target.
Skills + tools to load
maf://guide — the 1.3.0 migration guide; reverse every "after" → "before"
.github/skills/maf-obsolete-api-registry/SKILL.md — registry entries map 1.3.0 → 1.2.0 directionally
.github/skills/maf-migration-plan-creator/SKILL.md — tracking-table template (you'll mirror it)
Rollback Workflow
Phase A — Inventory the migration surface
Call MafDiffPackage("Microsoft.Agents.AI", "1.2.0", "1.3.0") to surface every breaking change between the two versions. The diff is the inventory — every entry is a potential rollback row.
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.
- 1mo ago First seen · 95 lines · 79 tokens per session scan A da9f455e1d96
MAF Rollback Agent is an agent published in the GitHub repository joslat/maf-doctor (14 stars, last pushed 4d ago), licensed MIT. It adds 79 tokens to every session and 1,596 once invoked, about $0.0003 per session on Opus 5.5. 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 2026-08-30.
Other agents, from other repositories
contrarian
Independently challenges a proposed Diagnostician root cause, trigger evidence, alternatives, confidence, fix strength, and diagram consistency.
diagnostician
Diagnose native Windows application, service, user-mode driver (including UMDF), and kernel-mode driver failures with public WinDbg workflows.
code-searcher
A read-only code-search agent for Java projects, especially multi-module Maven repositories. A monorepo is one repository containing several related modules or projects.
debugger
A debugging agent for Java applications built with Maven, a tool that manages Java builds and dependencies. It investigates build failures, registration problems, parsing issues, scanning errors, and MCP integration faults.
Error Handling Reviewer
Review error handling patterns — exception hierarchy, empty catch blocks, error boundaries, ProblemDetails consistency, retry logic, and user-facing error messages.
Performance Analyzer
Analyze performance issues: N+1 queries, missing caching, sync-over-async, allocation hotspots, missing indexes.