deepseek-harness

Let the OpenCode host delegate code review, development, and research tasks to DeepSeek Harness.

Quality

68%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./examples/plugin-packages/deepseek-harness/skills/deepseek-harness/SKILL.md

Quality

Content

82%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The body is a lean, instruction-only skill with unusually concrete API detail — exact endpoints, status enums, error codes, and gating conditions — plus validation checkpoints for service readiness, failure handling, and patch review. The main gaps are the missing invocation syntax for the extension call and the absence of section headers organizing the mixed setup, workflow, and auth rules.

Dimension	Reasoning	Score
Conciseness	The ~25-line body is dense with only operationally necessary, API-specific rules (capability gating, status enums, error codes, auth options, patch paging) and never explains concepts Claude already knows or pads with fluff — every token earns its place.	5 / 5
Actionability	Guidance is highly concrete — exact tool names (`capabilities`, `start`, `status`, `review`, `code`, `standard`, `cancel`), exact field values (`available: true`, `serviceStatus: "ready"`), error codes (`DSH_SERVICE_UNRESPONSIVE`), and parameters (`patchOffset`, `capabilities.runtimeManagement.supported`) — but the actual invocation syntax for `ipollowork_extension_call` (arguments, payload shape) is never shown, leaving a minor gap.	4 / 5
Workflow Clarity	The numbered rules form a clear sequence (capabilities gate → start → save jobId → poll status → handle failure → review patch before applying) with explicit validation checkpoints and error-recovery handling, fitting 'clear sequence with most checkpoints present'; the flow is interleaved with routing, auth, and runtime-management rules rather than presented as one chronological workflow, which keeps it below the level-5 anchor.	4 / 5
Progressive Disclosure	The skill is under 50 lines with no bundle files (references/, scripts/, assets/ absent) and no nested references, so a flat single-file layout is appropriate; structure is good (intro, invocation, policy, numbered rules, capability caveat) but there are no section headers separating setup, operational flow, and auth concerns, leaving minor organization gaps.	4 / 5
	Total	17 / 20 Passed

Description

53%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description has a clear, specific purpose tied to a named integration and reads in third person without fluff, but it entirely lacks a 'when to use' trigger clause and offers only generic task categories as keywords. Adding explicit trigger phrases and a broader set of natural synonyms would materially improve it.

Suggestions

Add an explicit trigger clause, e.g. "Use when the user asks for a second opinion, wants tasks delegated to DeepSeek/DSH, or requests heavier code review or research the main agent should offload."

Broaden trigger term coverage with natural synonyms users would actually say ("second opinion", "DSH", "DeepSeek", "offload this review") rather than only the generic categories "code review, development, research".

Consider naming one or two more concrete capabilities from the body (architecture analysis, patch generation with review-before-apply) to raise specificity beyond generic task categories.

Dimension	Reasoning	Score
Specificity	The description names its domain (delegating to DeepSeek Harness via the OpenCode host) and lists three task categories ("code review, development, and research tasks"), but uses a single generic verb ("delegate") rather than several concrete operations, matching the 'names domain and 1-2 concrete actions, but not comprehensive' anchor better than the level-4 'lists several specific actions' example.	3 / 5
Completeness	It clearly answers 'what' (delegate code review, development, and research tasks to DeepSeek Harness) but contains no 'when' clause or equivalent explicit trigger guidance, which per the judging guidelines caps completeness at 3 — the anchor for a clear 'what' with 'when' missing or only weakly implied.	3 / 5
Trigger Term Quality	Terms like "code review" and "research" are natural phrases users would say, but the description provides no "Use when..." trigger phrasing and misses common variations (e.g., 'second opinion', 'ask DSH', 'DeepSeek'), fitting 'some relevant keywords but missing common variations or synonyms'.	3 / 5
Distinctiveness Conflict Risk	Naming the specific "DeepSeek Harness" integration and the OpenCode host establishes a clear niche, but broad trigger words like "code review" and "development" overlap with ordinary review/dev requests where this skill should not fire, so it is 'mostly distinct; minor overlap risk' rather than the level-5 minimal-conflict anchor.	4 / 5
	Total	13 / 20 Passed

Validation

100%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository: Devin-AXIS/iPolloWork
Path: examples/plugin-packages/deepseek-harness/skills/deepseek-harness/SKILL.md
Commit: dada31d

Reviewed: 12 hours ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.