← DeepSeek Harness Plugins
DeepSeek Harness PluginManifest Valid

dsh-llm-verifier

Configurable DSH-native LLM verifier providing independent arbitration, pairwise comparison, and session acceptance gating.

UI & ProductivityDeveloper ToolsSecurity & PolicySkills & Workflows
19GitHub Stars1ForksUpdated2026-09-26

Install

$ dsh plugin --profile web add dsh-llm-verifier

Plugin Overview & Capabilities

AI-assisted organization based on the public repository snapshot. The content must be grounded in source evidence and does not replace compatibility or security verification.

source-grounded
dsh-llm-verifier introduces an independent judge and arbitration mechanism to DeepSeek Harness. While the primary Agent executes tasks, the verifier evaluates task objectives, candidate execution traces, and terminal outputs against configurable A–T rubrics. It includes five core tools (verifier_compare, verifier_select, verifier_track, verifier_best_of_n, verifier_current_session), multi-judge ensemble support, automatic routing, process selection, and persistent session caching.

Key Capabilities

  • Execute pairwise candidate comparison with A-T granular scoring and position bias mitigation
  • Run probabilistic pivot tournaments for multi-candidate evaluation (Best-of-N / verifier_select)
  • Track long-task and workflow completion progress using milestone checkpoint scoring
  • Perform automated host-level session acceptance gating on agent turn-stopping boundaries
  • Support multi-judge ensembles with median aggregation, transparent per-judge breakdown, and fail-soft degradation
  • Support process selection on llm/stream streams with recovery and every-step modes
  • Persist scoring caches, capability memory, and token usage statistics per DSH session topic

Useful For

  • Arbitrating between competing subagent/workflow execution strategies before committing changes
  • Enforcing strict automated acceptance gates on coding and debugging tasks based on real execution outputs
  • Tracking task progress in multi-step agent plans to detect regressions and incomplete deliverables
  • Benchmarking and ranking draft solutions using independent LLM judges with logprobs or explicit tag scoring

Who It Fits

  • Developers running complex multi-step workflows and autonomous coding agents in DeepSeek Harness
  • Engineers seeking deterministic verification, automated acceptance checks, and benchmark tournaments for AI outputs

Documented Limitations

  • Top logprobs expected scoring depends on provider/model compatibility; unsupported models smoothly fall back to explicit A-T tags
  • Automated routing and multi-judge evaluations consume additional LLM API call budgets and token quotas
  • Diff and file change evidence inspection requires DSH host version 0.1.6 or newer

DSH Compatibility

Version-specific runtime evidence collected by DSH Plugin. A missing result means we have not tested that combination yet.

Not tested yetNo runtime compatibility tests have been published yet.

Security Signals

Objective signals discovered from package metadata and source inspection. These are not a guarantee that a plugin is safe.

Dependency Counts

package.json declares 1 runtime, 22 development, 9 peer, and 0 optional dependencies.

info
Dsh Bundle Declared

package.json declares DSH bundle metadata.

info
License Declared

GitHub reports the repository license as MIT.

info
Package Manifest Available

A root package.json was captured and can be inspected by the registry.

info
Source Available

Public GitHub source metadata is available for this registry snapshot.

info

Source & Registry Notes

Public provenance, Registry classification, and the latest source check for this entry, kept separate from runtime verification.

Source repository
Aa728848/dsh-llm-verifier
Registry source
GitHub · dsh-plugin topic
Registry classification
Plugin
Source checked
b57158f · 2026-09-26

This project is independently indexed from public source information. DSH Plugin is not affiliated with DeepSeek or the plugin author. Always check the author repository before installation.

Repository Activity

GitHub stars
19GitHub stars
Forks
1
Open issues
0
Last commit
2026-09-26
Last release
No release detected
For maintainers

Maintaining this plugin?

This listing is generated from public repository data. If you maintain this project, you can review the information and share this listing with your users if you find it useful.

Add to README
Listed on DSHPlugin.app