dsh-auto-mode
DeepSeek Harness preset where Jev answers an agent's open end-of-turn questions, steering it on only above confidence and safety thresholds.
This summary is based on how curated lists describe the project; its README was not available to review.
Overview
dsh-auto-mode is a permission preset for DeepSeek Harness. According to the list summary, at the end-prompt step Jev answers the open questions an agent leaves in its final message, so the agent can keep working autonomously instead of waiting for the human.
It steers the agent back to work only when a choice clears 0.6 confidence and an autonomy-safety Noul clears 0.5; otherwise the turn returns to the human. It is hosted on a personal Git server and no README was available, so installation and evaluation details are unknown.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which answer should be given to the agent's open question? | — |
| Yes / no | Is it safe to continue autonomously with this answer? | yes · no |
How it works
- Agent ends a turn with open questions
- Jev picks answers and judges autonomy safety
- If choice ≥ 0.6 and safety ≥ 0.5, steer the agent back
- Otherwise return the turn to the human
Strengths
- Fewer interruptions for routine questions
- Explicit safety threshold before autonomy
Caveats
- List-only; no README reviewed
- Hosted outside GitHub
At a glance
- Type
- Article
- Status
- prototype
- Runtime
- DeepSeek Harness preset
- Has code
- yes
Use cases and integrations
How curated lists describe it
Coding agents: DeepSeek Harness permission preset whose end-prompt step has Jev answer the open questions an agent leaves in its final message, steering them back only when a choice clears 0.6 confidence and an autonomy-safety Noul clears 0.5, and returning the turn to the human otherwise.
yibie/awesome-jev
Frequently asked questions
What does Jev decide?
Answers to the agent's open questions and whether continuing autonomously is safe.
What are the thresholds?
0.6 confidence on the choice and 0.5 on the safety Noul, per the summary.
What happens below threshold?
The turn goes back to the human.
Related projects
- SystemOneHarness★ 193
Agent harness that compiles an environment's finite actions into typed questions for Jev and gates each step by confidence.
CLIbetaPython Decision engine that combines Jev probability distributions with loss matrices to decide when a clarifying question is worth asking.
ProjectprototypeTypeScriptMinimal agent loop where Jev selects each action and approves tool calls, a chat model fills arguments, and MCP tools execute.
MCP serverresearchPythonZero-dependency agent harness that sends each loop fork (act?, which tool, risk, done?) to a decision model and keeps the LLM for writing.
ServiceprototypeJavaScriptAgent skill that sends closed coding-agent judgments to Jev before asking the user, before risky commands and after failures.
ModelbetaPythonDeepSeek Harness plugin with 12 opt-in Jev or Luna Decisions judgments: skill and file selection, supervision, filtering and approval help.
PluginbetaTypeScript