DeepSeek Harness PluginManifest Valid

dsh-baize-rules

Injects persistent global, project, and session-level behavioral rules and reminders into DeepSeek Harness conversations.

UI & ProductivityTerminal & TUIDeveloper ToolsBrowser & Web

4GitHub Stars0ForksUpdated2026-09-22

Source Homepage

Install

$ dsh plugin --profile web add [email protected]

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-baize-rules (Baize) is a DeepSeek Harness plugin designed to inject user-defined 'must-do' and 'must-not' behavioral rules into model interactions. It supports three distinct rule scopes—global, session, and project—rendered in order of specificity (project > session > global) within a <system-reminder> frame. The plugin includes a slash command (/baize-rules), a dedicated web UI panel tab, tag-based filtering, template management with JSON import/export, and conflict-safe persistent storage.

Key Capabilities

Injects scoped rules (global, session, project) into conversation prompts at startup and regular intervals
Provides /baize-rules slash commands for full rule and template CRUD operations
Features an interactive Web UI panel integrated into conversation view slots
Supports rule templates with tag categorization and JSON export/import
Implements specificity-based budget trimming and SHA-1 deduplication to optimize model context
Ensures concurrent-safe writes and fault-tolerant reads across persistent rule files

Useful For

Enforcing personal coding guidelines (e.g., commenting languages, package manager preferences) across all sessions
Setting project-specific constraints and execution rules bound to the repository working directory
Managing reusable prompt templates and behavioral guardrails for different AI workflows

Who It Fits

Developers using DeepSeek Harness who need persistent behavioral guardrails across chats
Teams requiring project-level or session-specific coding conventions for AI agents
Power users managing extensive collections of custom prompt rules and templates

Documented Limitations

Web UI panel is accessible only from active conversation tabs; there is no standalone entry point on empty new-chat pages
Requires explicit schema configuration for default scope and maxBytes
Project scope functionality depends on the session having a valid working directory (cwd)

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 0 runtime, 15 development, 14 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: bvcvb/dsh-baize-rules
Registry source: GitHub · dsh-plugin topic
Registry classification: Plugin
Source checked: a7aae0f · 2026-10-10

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.