Skip to main content
Glama
xuanmingguo

DOT DeepSeek Harness MCP

by xuanmingguo
OverviewSchemaSourceRelated ServersDiscussionsAdmin
TypeScript
Local

Continue a Harness session

reply_run
Destructive

Submits a new queued turn to a completed run session, optionally selecting a different model, and tracks it as a new run.

Instructions

Submit a new queued turn to the completed run session, optionally selecting a different model, and track it as a new run. Task parameters are path-reference-only: provide the authorized workspace, reviewTargets, contextReadScope, excludedPaths, writeScope, and acceptance criteria. reviewTargets identify what to assess; they are not a read whitelist. contextReadScope declares where Harness may search and read supporting implementation, tests, configuration, and architecture material. Never embed source text, diffs, file dumps, encoded source, or repository archives. Harness reads named files from the authorized workspace itself. These fields are task instructions, not enforced per-path filesystem isolation. The selected Harness model may send content it reads to its configured model provider even though Relay itself uses a loopback address.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
taskNoTask parameters are path-reference-only: provide the authorized workspace, reviewTargets, contextReadScope, excludedPaths, writeScope, and acceptance criteria. reviewTargets identify what to assess; they are not a read whitelist. contextReadScope declares where Harness may search and read supporting implementation, tests, configuration, and architecture material. Never embed source text, diffs, file dumps, encoded source, or repository archives. Harness reads named files from the authorized workspace itself. These fields are task instructions, not enforced per-path filesystem isolation. The selected Harness model may send content it reads to its configured model provider even though Relay itself uses a loopback address.
modelNo
runIdYes
contentNoTask parameters are path-reference-only: provide the authorized workspace, reviewTargets, contextReadScope, excludedPaths, writeScope, and acceptance criteria. reviewTargets identify what to assess; they are not a read whitelist. contextReadScope declares where Harness may search and read supporting implementation, tests, configuration, and architecture material. Never embed source text, diffs, file dumps, encoded source, or repository archives. Harness reads named files from the authorized workspace itself. These fields are task instructions, not enforced per-path filesystem isolation. The selected Harness model may send content it reads to its configured model provider even though Relay itself uses a loopback address.
providerNo
writeScopeNoPaths Harness may modify. Must be empty for read-only runs; this is an instruction declaration in addition to native permissions.
openBrowserNoKeep false unless the user explicitly asks to open the Harness page in the OS browser.
excludedPathsNoPaths excluded from contextual reading. This is an instruction declaration, not filesystem enforcement.
externalPathsNoExplicit external read/write file or directory grants; operator policy must separately allow each. External writes require danger-full-access.
reviewTargetsNoSubjects to assess. This is not a read whitelist; supply together with contextReadScope.
idempotencyKeyNo
reasoningEffortNo
contextReadScopeNoWorkspace paths Harness may search and read for evidence. Use only the targets when the user explicitly requests a target-only review.
permissionPresetNoOmitted: inherit the parent run permissionPreset, regardless of the start_run configured default or the session current/restored preset. Explicit value overrides inheritance.
authorizationBasisNoTruthful evidence that the user explicitly selected Harness or this named Harness model to inspect the stated workspace scope. The current explicit request is sufficient authorization for that selected provider to process in-scope reads; do not ask the user to repeat authorization solely because provider processing is external. This field does not broaden the workspace, scope, permissions, destination, or allowed external actions.
confirmedDangerousPermissionNoAlways defaults to false on reply_run, including trusted-operator deployments. An inherited or explicit danger-full-access preset requires true for this call.

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observedv0.2.18-dsh021a2.14

TDQS

B3.3/5.0
Behavior4/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations already declare destructiveHint=true, openWorldHint=true and idempotentHint=false, so the safety bar is lower. The description earns credit for disclosing two things annotations cannot: the session must be completed, and content read by the selected Harness model may be sent to an external model provider despite Relay's loopback address.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness3/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The purpose sentence is well front-loaded, but well over half the description duplicates the task parameter text from the schema, which is redundant in a description. Every sentence about scope is load-bearing for the task field, yet presenting it here repeats structured data rather than adding value.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness3/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a 16-parameter mutation tool with no output schema, the description covers the task-construction semantics well, yet omits the required runId, the danger-confirmation flow (confirmedDangerousPermission) and permissionPreset inheritance. Since there is no output schema, the description could at least sketch what is returned, but an agent can still call the tool from what is given.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 69%, and the bulk of the description is a near-verbatim copy of the task/content schema description, so it adds little beyond what the schema already states. It does enumerate the scope fields expected inside the task string, but parameters like idempotencyKey, model, provider, reasoningEffort, permissionPreset and confirmedDangerousPermission are untouched.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

The opening sentence gives a specific verb, resource and scope: submit a new queued turn to the completed run session, optionally with a different model, tracked as a new run. This distinguishes it from siblings like start_run and queue_run by anchoring on an existing completed session, but it never names those alternatives.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines2/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

There is no explicit when-to-use guidance and no routing against siblings such as queue_run, steer_run, or retry_run_with_fallback. The closest thing to a condition is 'completed run session,' which is implied pre-state rather than usage instruction.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Deploy Server

Other Tools