dsh-rule-engine
A DeepSeek Harness rule-execution plugin that parses ~/.dsh/AGENTS.md and applies configurable guards, checks, and audit logging.
Install
$ dsh plugin --profile web add dsh-rule-enginePlugin 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.
Key Capabilities
- Parses rules from ~/.dsh/AGENTS.md and reloads them when the file changes
- Guards change-oriented tool operations while allowing listed read-only operations
- Provides /guard commands for status, rules, audit logs, reload, bypass, unlock, and authorization revocation
- Records audit events in ~/.dsh/rule-engine.log.jsonl
- Supports rule confidence levels, with low-confidence rules excluded from hard blocking
- Supports optional task-contract and anti-overengineering controls, including observation and armed modes
- Integrates optional LLM-assisted understanding for non-high-confidence rules with pattern-library fallback
- Reads disabled-rule state used by the companion dsh-rules-manager plugin
Useful For
- Enforcing user-maintained AGENTS.md operational rules in DSH sessions
- Auditing sensitive file changes, command activity, and authorization-related actions
- Applying temporary guard bypasses or configuration unlocks through explicit /guard commands
- Reviewing which rules were active and why a guard decision occurred
Who It Fits
- DeepSeek Harness users who maintain rules in AGENTS.md
- Users seeking mechanism-level enforcement and auditing of DSH workflow rules
- Users of the related dsh-rules-manager disabled-rule workflow
Documented Limitations
- The project describes itself as runnable but still requiring continued refinement, with possible incompleteness, false positives, and edge cases.
- Several process-oriented rules are currently primarily self-verification prompts rather than machine-decidable enforcement.
- Authorization and backup records are in memory and are lost after restart.
- DSH cannot block assistant/message output in real time; the plugin can only audit and inject corrections afterward.
- LLM understanding currently runs once for non-high-confidence rules and falls back to a pattern library if it fails.
- Requires Node.js >=22 and declares DSH compatibility from 0.1.0-rc.3 up to, but not including, 0.2.0.
DSH Compatibility
Version-specific runtime evidence collected by DSH Plugin. A missing result means we have not tested that combination yet.
Security Signals
Objective signals discovered from package metadata and source inspection. These are not a guarantee that a plugin is safe.
Source & Registry Notes
Public provenance, Registry classification, and the latest source check for this entry, kept separate from runtime verification.
- Source repository
- jilian-dsh/dsh-rule-engine
- Registry source
- GitHub · dsh-plugin topic
- Registry classification
- Plugin
- Source checked
- 490e52f · 2026-09-21
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
- 3GitHub stars
- Forks
- 0
- Open issues
- 1
- Last commit
- 2026-09-20
- Last release
- 2026-09-20
Related DSH Plugins
Ranked by overlapping capabilities, use cases, plugin type, categories, and DSH profile.
Persistent multi-model Agent teams and bounded DAG workflows with GUI controls for DeepSeek Harness.
View plugin →Local-first AI agent runtime and DSH plugin bundle providing sandboxed sessions, MCP tools, audit logs, and session replay.
View plugin →Second-model AI auto-review for DeepSeek Harness approval requests with fail-closed safety and session audit.
View plugin →Composable three-tier memory control plane for DeepSeek Harness with runtime context, searchable documents, and pluggable long-term memory.
View plugin →