AutoSaddler is a system that improves LLM-agent harnesses by diagnosing execution traces and applying structured changes to prompts, tools, middleware, and agent-loop logic. It evaluates candidate updates for their ability to generalize beyond the traces that motivated them. The catalogue add-ons represent workflows for using AutoSaddler.
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 microsoft/AutoSaddler --skill patch-verificationgit clone --depth 1 https://github.com/microsoft/AutoSaddlerWrote 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/microsoft/autosaddler/patch-verification)<a href="https://agentmods.dev/skills/microsoft/autosaddler/patch-verification"><img src="https://agentmods.dev/badge/skills/microsoft/autosaddler/patch-verification/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/microsoft/autosaddler/patch-verification"><img src="https://agentmods.dev/badge/skills/microsoft/autosaddler/patch-verification.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00038 | $0.00974 |
| Opus 5.5 | $0.00015 | $0.00390 |
| Sonnet 5.5 | $0.00008 | $0.00195 |
| Haiku 4.5 | $0.00004 | $0.00097 |
Grade A, and why
patch-verification 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 — 126 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Patch Verification
Overview
A patch that crashes or introduces bugs at runtime is worse than no patch at all. Every patch must pass thorough verification before being committed.
Only fix what would actually crash or produce incorrect behavior — do NOT change the patch's intent.
When to Use
After applying ANY patch type (capability or steering).
Verification Procedure
Step 1: Review the Diff
git diff
- Does the diff match your intent? No unintended changes?
- Are there leftover debug statements, commented-out code, or TODOs?
- Are there any files modified that should not have been changed?
Step 2: Syntax and Import Check
For every modified Python file:
python3 -m py_compile <modified_file>
python3 -c "import <modified_module>"
- No
SyntaxError,IndentationError, orTabError - No
ImportErrororModuleNotFoundError - No
NameErrorfrom undefined variables or functions at module level
Step 3: Docstring Integrity
Docstrings are the agent's only interface to tools — errors here directly cause agent failures.
- No placeholder values (
[...],{...},<placeholder>) — these crash downstream parsers - All examples use concrete, valid values
- Parameter names in the docstring match the actual method signature
- Return value description matches what the implementation actually returns
- No stale documentation from copy-paste (e.g., wrong tool name, wrong parameter list from another method)
- No literal escape sequences (
\n,\t, etc.) used to represent multi-line examples or formatting inside docstrings — these produce actual control characters in the parsed string and can causeSyntaxErroror silently corrupt the docstring text. Use triple-quoted strings with real newlines instead
Step 4: Template String Safety
If the patch modifies any string that uses .format() or f-string
interpolation:
- Every
{placeholder}has a corresponding.format()argument - No extra or missing placeholders from copy-paste
- No unescaped curly braces in literal text (use
{{and}}to escape) - No cross-file contamination between template variables
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 · 126 lines · 38 tokens per session scan A 38518d33a4ba
patch-verification is a skill published in the GitHub repository microsoft/AutoSaddler (243 stars, last pushed 3d ago), licensed MIT. It adds 38 tokens to every session and 974 once invoked, about $0.0002 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 skills, from other repositories
research-engineer
An uncompromising Academic Research Engineer. Operates with absolute scientific rigor, objective criticism, and zero flair. Focuses on theoretical correctness, formal verification, and optimal implementation across any required technology.
tika-eval-compare
Compare extracts from two Tika builds over a corpus to detect regressions in content, encoding, exceptions, and embedded-document handling. Use for "compare before/after extracts", "eval this change against the corpus".
cua-driver
Use Cua Driver for desktop or browser tasks that are awkward or unavailable through Bash/APIs, or when the user explicitly wants GUI interaction: app testing, visual bug reproduction, form filling, calendar entry, screenshots, and demo recording. Also covers Cua setup; not OpenAI Codex Computer Use or web research.
jetson-validate-image
Use after jetson-flash-image to run static BSP checks, on-target smoke/regression tests on a flashed DUT, or both. Not for build or flash steps. Triggers: validate bsp, on-target validation.
atmos-validation
Validate Atmos projects, components, arbitrary JSON Schema inputs, EditorConfig, and GitHub Actions; use affected-file selection and native CI annotations.
skill-benchmark
Benchmark AI skill effectiveness by measuring implementation quality against legacy constraints.