Skills/skills.sh/microsoft-foundry

microsoft-foundry@r7

VERIFIED· install only

Publisher claims: "Build, deploy, evaluate, optimize, fine-tune, and manage Microsoft Foundry agents, models, and resources end to end. USE FOR: foundry, azd ai agent, azd provision/deploy, hosted agent scaffold/develop/run/deploy/troubleshoot, prompt agent, managed harness agent, Copilot harness, voice agent, create/update/invoke/evaluate agent, add tool to agent, agent insights, batch/continuous eval, continuous monitoring, agent CI/CD, optimize/improve prompt, optimize agent instructions, Agent Optimizer, dataset curation from traces, deploy model, model fine-tuning (SFT/DPO/RFT), Foundry project, RBAC, role assignment, permissions, quota, capacity, region, deployment failure, AI Services, create Foundry resource, knowledge index, customize deployment, onboard, availability, training-data, grader, distillation, large file upload. DO NOT USE FOR: Azure Functions, App Service, general Azure deploy (use azure-deploy), general Azure prep (use azure-prepare)." Observed in sandbox: no network, touches nothing outside the project, writes no memory.

Registry skills.shPublisher microsoft · publishing for 1198 daysInstalls 736KLast detonated 14 min ago
$npx skills add microsoft-foundry
WatchReport an issue

Coverage

Bundled scripts ran with no arguments; no agent task scenario for this bundle.
Tested
8 of 15 bundled scripts run with no arguments (the rest were not run: per-bundle cap), with decoy credentials in the environment · every file read, file written and host contacted by what ran
Not tested
use of the skill by an agent

What changed since r1

vet-skills diff microsoft-foundry r1 r7

This version reaches for something r1 did not: 1 new host (azure.com) · 2 new credential variables (API_KEY, AZURE_OPENAI_API_KEY). Compared from the two recorded runs, not from source.

New hosts
azure.com
New credential variables
API_KEY, AZURE_OPENAI_API_KEY

r1 was not run in the sandbox (static only); the comparison covers what static analysis recorded only.

The guard can hold an update like this until someone approves it (runtime.on_change: hold). A change that only shows with a real credential, or on a path the run did not reach, is not in a trace and so not in this comparison.

Is microsoft-foundry@r7 safe to install?

KillerSkills ran it in the KillerSkills honeypot sandbox (2 scenarios; bundled scripts ran with no arguments; no agent task scenario for this bundle) on 2026-11-17 and scored it 90/100 (Verified). In what ran, nothing we could detect went wrong: no undeclared network egress, no credential reads, no persistent instruction writes. What it does with a real credential was not observed; the static findings and the publisher carry the rest of the score. Summary: no network, touches nothing outside the project, writes no memory.

Behavior trace

Honeypot run · 1 s · 2 scenarios · 0 deviations
t+CategoryObserved action
0.0sprocessno install scripts in bundle
0.0sprocessbundled script run in the sandbox with no arguments
0.1sprocessbundled script run in the sandbox with no arguments
0.1sprocessexited 2: error: the following arguments are required: --job-id
0.1sprocessbundled script run in the sandbox with no arguments
0.2sprocessbundled script run in the sandbox with no arguments · not named in the instructions
0.2sprocessbundled script run in the sandbox with no arguments
0.3sprocessexited 2: error: the following arguments are required: --input, --output, --format
0.3sprocessbundled script run in the sandbox with no arguments
0.4sprocessexited 1: ModuleNotFoundError: No module named 'requests'
0.4sprocessbundled script run in the sandbox with no arguments
0.5sprocessexited 2: error: the following arguments are required: --deployment-name, --test-file
0.5sprocessbundled script run in the sandbox with no arguments
0.5sprocessexited 1: ModuleNotFoundError: No module named 'openai'
0.5sprocess8 of 15 bundled scripts run with no arguments (the rest were not run: per-bundle cap)
0.5sprocessscript backend runs setup only
0.5sprocessscript backend runs setup only
0.5sprocessno install scripts to re-run

Static findings

MEDIUM
Python subprocess execution detectedfinetuning/scripts/deploy_model.py:75 · Python subprocess execution detectedresult = subprocess.run(Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
File type combined with upload actionfinetuning/scripts/evaluate_model.py:196 · File type combined with upload action — agent may exfiltrate local data filesparser.add_argument("--output", default="eval_results.json", help="Output file")Fix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST07
MEDIUM
Python exec or eval of computed codefinetuning/scripts/calibrate_grader.py:54 · Python exec or eval of computed codeexec(compile(source, grader_path, "exec"), namespace)Fix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST05
LOW
Prompt injectionfoundry-agent/create/references/guardrails/guardrail-manage.md:56 · Prompt injection — known jailbreak technique reference · informational: not an instruction file| `Microsoft.DefaultV2` | Updated default with jailbreak + protected material | No |Fix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.
AST01
LOW
Prompt injectionfoundry-agent/create/references/guardrails/guardrail-api-create.md:42 · Prompt injection — known jailbreak technique reference · informational: not an instruction file{ "name": "Jailbreak", "enabled": true, "blocking": true, "source": "Prompt" }Fix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.
AST01

Cost to the agent

Instructions loaded
≈8,989 tok
2 files · 60 in bundle

Tokens estimated at four characters each. Skills load their instruction text when triggered; the figure is what that costs.

Known vulnerabilities

Vulnerability advisories (OSV) cover npm and PyPI packages; this registry has no advisory database.