Diagnosis · Writing

docs-grounding-verifier

Name: docs-grounding-verifier
Author: microsoft

Agent Package Manager

From Microsoftgithub.com/microsoft/apm ↗Source ↗MIT3,616 repo stars

Checking the letterforms and the sentence

Add it to your AI

View source

Works with

ClaudeChatGPTGeminiClaude CodeCodexCursorCopilotGemini CLI

Setup

One command

Best in a coding tool, where its scripts can run.

Contains

Instructions + 4 scripts

Shell and Python.

Maintained

Last change 3 Oct 2026

On the repository’s main branch.

Check

1 to review

What it does

CLAIM-LEVEL grounding verification. Adapts the RAGAS faithfulness-eval pattern (proven in RAG literature) to docs/code instead of generated- answers/retrieved-context. Source code is the ground truth; docs paragraphs are the candidate text under audit.

What it can touch

Includes 4 scripts
Your AI can run these when the skill is used — on your computer in a coding tool (which asks first unless you have told it not to), or in the AI’s own sandbox in Claude and ChatGPT.
How we know
Found in the skill’s folder: evals/run-evals.sh, scripts/extract-claims.py, scripts/retrieve-evidence.sh, scripts/verify-page.sh.
4 smaller things worth knowing
Sets itself up to run again automatically later; Changes your AI's own settings, permissions or connectors. Usually legitimate — listed so nothing is hidden.
How we know
Sets itself up to run again automatically later — evals/runs/proof/pkgtypes/evidence/docs_src_content_docs_reference_package-types_md_c1.json line 1; Changes your AI's own settings, permissions or connectors — evals/runs/proof/pkgtypes/evidence/docs_src_content_docs_reference_package-types_md_c12.json line 1; Sets itself up to run again automatically later — evals/runs/proof/pkgtypes/evidence/docs_src_content_docs_reference_package-types_md_c13.json line 1; Sets itself up to run again automatically later — evals/runs/proof/pkgtypes/judge-batch-docs_src_content_docs_reference_package-types_md.txt line 116. Read 2026-10-05, commit 18c4c43.
evals/runs/proof/pkgtypes/evidence/docs_src_content_docs_reference_package-types_md_c1.json, line 1 ↗evals/runs/proof/pkgtypes/evidence/docs_src_content_docs_reference_package-types_md_c12.json, line 1 ↗evals/runs/proof/pkgtypes/evidence/docs_src_content_docs_reference_package-types_md_c13.json, line 1 ↗evals/runs/proof/pkgtypes/judge-batch-docs_src_content_docs_reference_package-types_md.txt, line 116 ↗
No serious risk pattern found
Launch7’s automated check read all 101 text files in this skill and matched none of its serious risk patterns. This is not a guarantee: a check like this can miss new or disguised tricks, and the author can change the files after the check.
How we know
Read 2026-10-05, commit 18c4c43 by pipeline/trust/analyze.py — 40+ patterns for hidden text, prompt injection, downloads that run, credential access and more. It reads; it never runs the skill.
Recently maintained
Its repository has had changes recently.
How we know
Last push to microsoft/apm: 2026-10-03 (2 days before the scan).

Read automatically from the skill’s own files on GitHub on 5 Oct 2026. Launch7 does not run or manually review skills: this lists what the check saw, and nothing here means “safe”. How we check

Add it to your AI

Partly. Upload it in Claude’s settings — the instructions work; the scripts need the full folder.

Built in your browser from the SKILL.md on GitHub — nothing passes through Launch7.
In Claude, open Customize → Skills, press +, then Create skill → Upload a skill, and choose the .zip.
Make sure code execution is switched on (Settings → Capabilities). Skills need it.
Start a new chat and ask for the job in your own words — Claude picks the skill up when it fits.

This skill ships 4 scripts beside its instructions. The download here is the SKILL.md alone — enough for the instructions, not for the parts that run code. Use the command route, or take the full folder from the repository.

How Claude does this: support.claude.com · checked 28 Sep 2026 · “Supported” means documented by the vendor; Launch7 has not run it end to end

Where it comes from

Publisher: microsoft
Found on: GitHub
Licence: MIT
Source dated: 23 Aug 2026
Repo stars: 3,616

Stars count the whole repository, which holds 12 other listings here.

Why it is listedThis ships from github.com/microsoft, which is Microsoft's own account. GitHub controls who can publish under an account name, so the source is Microsoft rather than someone naming them.

Similar

source-verification
Verify sources, claims, images, video, documents, interviews, and synthetic media with SIFT and a durable evidence trail.
jamditisIts code sends data over the internet
spec-drift-check
Audits a WP brief for drift against current code — extracts claims, verifies each, emits PROCEED/STOP verdict.
QBall-IncInstructions only
module-level-code-translator
Translate source code between programming languages at function, class, and module levels while preserving behavior and generating verification tests.
ArabelaTsoIncludes 4 scripts
verification-before-completion
Use when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before assertions a
obraInstructions only
doublecheck
Three-layer verification pipeline for AI output.
From GitHubInstructions only

What it does

What it can touch

Add it to your AI

More from microsoft