dsh-model-fixer

PluginAI Models

Verification

L1 · Found

Risk

Low

Health

Active

Trust

Unrated

What it doesAI

Fixes sandbox escalation loops in DeepSeek Harness by stripping invalid sandbox params and adding model-specific protocol prompts.

Strips invalid sandbox permissions/justification from tool calls
Model-agnostic: auto-applies to any model
Optional per-model protocol prompt, GPT by default

AI-generated from the repo README — for reference only.

Installation

dsh plugin --profile web add github:Marsax110/dsh-model-fixer

Install method: GitHub · no verification record (L1-L5)

Compatibility

DSH Version	Status
not stated	no verification record

Requirements

• Node.js: not stated
• DSH: not stated
• External credentials: none detected

Security Report

Automated static scan, not manual review.

DSH.SO VETPASSED93deb5

Security reviewAutomated review · daily

PASSED

Vet static analysis found no suspicious behavior — verdict is clean.

0 critical·0 high·0 medium·100 score

Version binding

plugin version0.1.0

scanned commit93deb505f6a32026-08-25

latest commit93deb505f6a3

vet verdictclean · gitscore 100

Vet is an AST static scan (npm artifact or git source).

Disclaimer: automated static analysis, not a security guarantee. Always review what you install.

Risk details

No critical or warning findings.

Activity

Last commit 2026-08-25 · activity: Active

• Repo created: 2026-08-25

• Stars: ★ 0 · Forks: 0

• Health: Active — committed within last 30 days

Source

GitHub: github.com/Marsax110/dsh-model-fixer

Plugin notes

What is the dsh-model-fixer DeepSeek Harness plugin?

dsh-model-fixer is a DeepSeek Harness plugin maintained by @Marsax110. It is cataloged on dsh.so as a AI Models plugin, with a public GitHub source, declared install command, and automated static security scan.

How do I install dsh-model-fixer?

Run dsh plugin --profile web add github:Marsax110/dsh-model-fixer in a DeepSeek Harness (dsh) environment. The command is parsed from the plugin manifest, but it has not yet been tested in an isolated sandbox at the L3+ level.

Is the security scan on this page a manual audit?

No. The scan is an automated heuristic review of the repository source and can produce false positives. It is not a manual security audit and never an endorsement; review the repository and installed permissions before using the plugin.

"DeepSeek Harness (dsh) is an open-source agent harness developed by DeepSeek AI. It uses an architecture where everything is a plugin, and is powered by Cordis."
— DeepSeek Harness README

Sources: GitHub repository · DeepSeek Harness README · security scan data · Maintained by dsh.so · Last updated 2026-08-25