CategoryWeightCategory scoreContribution

Securityprompt, exec, net, exfil, eval

35%

30.8 pts

Supply chainhash, typosquat, maintainer, lockfile

20%

100

20.0 pts

Maintenancestaleness, pinning, CI

15%

100

15.0 pts

TransparencySKILL.md, perms, README

15%

100

15.0 pts

Communityinstalls, verify, response

15%

100

15.0 pts

Findings & checks · 1 flagged

Securityscore 88 · 1 finding

MEDIUMInstruction telling the agent not to ask for approvalSS-SKILL-INJECT-DONT-ASK-01 · Prompt injection · README.md

MEDIUM — it fires on intent; the real damage depends on the host agent's own approval-gating.

Why it matters

The text {match} tells the agent to skip the normal "ask the user first" gate. Used adversarially it removes the human-in-the-loop check before destructive or sensitive actions, turning a normally-gated agent into a fire-and-forget executor.

The exact value spotted

excerptREADME.md· markdown

53The GitHub Copilot for Azure extension installs skills when the extension activates. At that

… (108 chars elided on L53)

55![Toast notification asking "GitHub Copilot for Azure can install skills to enhance your exp

… (108 chars elided on L55)

57To manually install skills from the Command Palette, the following commands are available:

Occurrences

1 occurrence · at L55

How to fix

Remove the approval-skipping instruction, or scope it narrowly to a specific safe, reversible action.

Delete blanket "don't ask / no need to confirm" directives from the skill.
If the skill is a genuine autonomous job, restrict the opt-out to a named non-destructive action rather than all actions.

Framework references

OWASPLLM01 ATLASAML.T0051

Trace & refs

ruleSS-SKILL-INJECT-DONT-ASK-01sha256916b0a531165cbd7…rubric 365aacaView on GitHub

Supply chainscore 100 · 0 findings

✓—All supply chain checks passedNo findings in this category for the latest scan.pass

Maintenancescore 100 · 0 findings

✓—All maintenance checks passedNo findings in this category for the latest scan.pass

Transparencyscore 100 · 0 findings

✓—All transparency checks passedNo findings in this category for the latest scan.pass

Communityscore 100 · 0 findings

✓—All community checks passedNo findings in this category for the latest scan.pass

The primary manifest — the file an agent reads to learn what this artifact does.

.github/skills/file-test-bug/SKILL.md · 1 file

.github/skills/file-test-bug/SKILL.md2.5 KB · Markdown

name: file-test-bug

description: "File a GitHub issue for local integration test failures. TRIGGERS: file test bug, report test failure, create bug for test, integration test failed, test failure issue, junit failure"

license: MIT

metadata:

author: Microsoft

version: "1.0.0"

File Test Bug

Creates a GitHub issue in microsoft/github-copilot-for-azure for integration test failures.

Input

Skill name (required): e.g., azure-ai, appinsights-instrumentation
Test run (optional): Timestamp of test run. Defaults to most recent with logs for the skill.

Steps

Ask user for skill name if not provided
Parse tests/reports/junit.xml for failures matching the skill
Find test run directory (specified or most recent with matching logs)
Read agent-metadata.md from tests/reports/test-run-<timestamp>/<skillname>-<testname>/
For each failure, read the actual line of code from the test file using the location (file:line) from junit.xml
REQUIRED - Write diagnosis BEFORE creating issue:

Analyze the agent-metadata.md to understand what the agent did
Compare agent behavior to what the test expected (from the assertion)
Identify the root cause (skill issue, test issue, or model behavior)
Write 2-3 sentences per failed test explaining WHY it failed
Suggest potential fixes (update skill, update test, or update fixtures)

Create issue via github-mcp-server-create_issue:

owner: microsoft
repo: github-copilot-for-azure
title: Integration test failure in <skill-name>
labels: ["bug", "integration-test"]
body: |
  ## Failed Tests
  - <test-name>: <error message>
  
  ## Diagnosis
  
  ### Root Cause
  <1-2 sentences explaining WHY the test failed based on agent-metadata.md analysis>
  
  ### Analysis per Test
  - **<test-name>**: <what agent did vs what test expected>
  
  ### Suggested Fix
  <one of: update skill, update test assertions, provide test fixtures>

  ## Details
  ### <test-name>
  **Error:** <failure from junit.xml>
  **Location:** <file:line>

  
  <details>
  <summary>agent-metadata.md</summary>
  
  <full contents of agent-metadata.md file, verbatim>
  
  </details>

Important

Include the complete, unmodified contents of each agent-metadata.md file in the issue body. Do NOT summarize or truncate the logs. Wrap each log in a <details> block with the test name as the summary.

file-test-bug— agent skill

Is file-test-bug safe to install?

File Test Bug

Input

Steps

Important

Audit the pieces. Scan the whole. Decide.