dsh-approval-llm
A DeepSeek Harness permission plugin that uses a separate reviewer model to approve, deny, or escalate tool-approval requests.
Install
$ dsh plugin --profile web add dsh-approval-llm # installs the published npm packagePlugin 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
- Adds a model-approval permission mode and preset for DeepSeek Harness.
- Routes requests through deterministic allowlist, denyList, and humanOnlyList rules before model review.
- Uses a reviewer model to return ALLOW, DENY, or ESCALATE decisions.
- Delegates reviewer timeouts, parse errors, and provider errors to human approval.
- Supports a per-session consecutive-denial circuit breaker.
- Can append user-visible model approval or denial notices to the session.
- Ships a bundled skill for configuring the approval reviewer.
Useful For
- Low-risk workflows where tool approvals can be reviewed by a separate model.
- Keeping selected sensitive tools exclusively under human approval.
- Automatically allowing or rejecting explicitly named tools through routing lists.
- Adding an AI-proposes, user-confirms configuration flow for approval review.
Who It Fits
- DeepSeek Harness users operating the web profile.
- Teams configuring permission review policies for agent tool calls.
- Developers who want a model-assisted approval layer while retaining human escalation.
Documented Limitations
- AI reviewer decisions are a deployment policy choice, not a security guarantee, and may be affected by prompt injection.
- Tool routing lists use exact tool-name matching only; wildcard and argument-pattern routing are not available.
- The plugin processes one approval request per reviewer model call; batch review is not implemented.
- It has no dedicated log-only audit event for the full review context.
- It ships no client-side tool-card badge yet.
- Explicit reviewer provider and model settings must be configured together; fallback routing depends on a prior session request header.
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.
package.json declares 0 runtime, 13 development, 10 peer, and 0 optional dependencies.
package.json declares DSH bundle metadata.
GitHub reports the repository license as MIT.
A root package.json was captured and can be inspected by the registry.
Public GitHub source metadata is available for this registry snapshot.
Source & Registry Notes
Public provenance, Registry classification, and the latest source check for this entry, kept separate from runtime verification.
- Source repository
- Letter2025/dsh-approval-llm
- Registry source
- GitHub · dsh-plugin topic
- Registry classification
- Plugin
- Source checked
- 0918a5a · 2026-09-14
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
- 7GitHub stars
- Forks
- 0
- Open issues
- 1
- Last commit
- 2026-09-08
- Last release
- No release detected
Related DSH Plugins
Ranked by overlapping capabilities, use cases, plugin type, categories, and DSH profile.
Second-model AI auto-review for DeepSeek Harness approval requests with fail-closed safety and session audit.
View plugin →Native SQLite-backed local project taskboard plugin and UI for DeepSeek Harness with Agent claim and review flows.
View plugin →Persistent multi-model Agent teams and bounded DAG workflows with GUI controls for DeepSeek Harness.
View plugin →Compatibility bridge and host ABI enabling unmodified Pi plugins to run natively on DeepSeek Harness.
View plugin →