What is deepseek-harness?

A local web interface for handing a scoped coding task to DeepSeek Harness, an agent environment where the work can be watched in a browser.

Starting a Harness session, passing it a defined coding task, and reviewing the workspace changes afterward.

It makes delegated coding work visible and keeps the main agent responsible for inspecting and verifying the resulting changes.

deepseek-harness — Skill for Claude Code and Codex by Seann0824

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent

npx skills add Seann0824/deepseek-harness-for-codex --skill deepseek-harness

Clone the repo

git clone --depth 1 https://github.com/Seann0824/deepseek-harness-for-codex

Made for: Claude Code, Codex.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

README.md

[![agentmods](https://agentmods.dev/badge/skills/seann0824/deepseek-harness-for-codex/deepseek-harness/github.svg)](https://agentmods.dev/skills/seann0824/deepseek-harness-for-codex/deepseek-harness)

Your own site

<a href="https://agentmods.dev/skills/seann0824/deepseek-harness-for-codex/deepseek-harness"><img src="https://agentmods.dev/badge/skills/seann0824/deepseek-harness-for-codex/deepseek-harness/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

Your own site · 80×15

<a href="https://agentmods.dev/skills/seann0824/deepseek-harness-for-codex/deepseek-harness"><img src="https://agentmods.dev/badge/skills/seann0824/deepseek-harness-for-codex/deepseek-harness.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 62 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 629 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

Origin original No closer match found in the catalogue.

Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

Model	Per session	Once invoked
Fable 5.1	$0.00062	$0.00629
Opus 5.5	$0.00025	$0.00252
Sonnet 5.5	$0.00012	$0.00126
Haiku 4.5	$0.00006	$0.00063

Measured 2mo ago against content hash f2183a413966, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

deepseek-harness scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2mo ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

plugins/deepseek-harness/skills/deepseek-harness/SKILL.md · 27 lines

How it starts

The opening of the file, as written. The whole thing — 27 lines — stays where its author put it; the contents beside it link to each section on GitHub.

DeepSeek Harness

Use the deepseek-harness MCP tools to run DeepSeek Harness locally through its Web UI. The user must be able to watch the same session that Codex controls. Codex remains responsible for the task outcome and must inspect the changed files and run appropriate verification itself.

Workflow

Call doctor before the first run in a task. Explain any failed prerequisite. A DeepSeek API key may come from the MCP environment, the target workspace's uncommitted .env file, or the Harness Web Models settings.
Inspect the target repository enough to write a concrete delegation prompt. Include the requested outcome, relevant repository instructions, scope limits, and acceptance checks. Do not include credentials.
Decide whether the attempt needs a new Harness session or should continue a completed one. Omit sessionId for independent work or a clean retry. Pass the earlier run's sessionId for a direct follow-up or correction that benefits from the existing conversation. Never reuse a running session or a session from another workspace.
Call start_run exactly once for that attempt with openBrowser: false and the chosen sessionId, if any. It starts or reuses dsh web and submits the task without opening a browser.
Show the returned webUrl as a clickable Markdown link so the user can open the live page when they choose. State whether sessionReused is true. Do not call open_service unless the user explicitly asks Codex to open the page.
Call wait_run with a timeout of at most 30 seconds until the run is succeeded, failed, or cancelled. The browser and MCP observe the same session ID and history.
Treat the Harness response as a handoff report, not proof. Inspect the actual workspace diff, preserve unrelated user changes, and run the smallest checks that cover the change.
If review finds a concrete defect, choose whether to continue the completed session for context or start a clean correction session, state the defect and failed check precisely, and review the correction again.
Use cancel_run to cancel an unsafe or obsolete turn while leaving the Web page available. Use stop_service only when the local server is no longer needed.

Read the full file on GitHub · 27 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

2mo ago First seen · 27 lines · 62 tokens per session scan A f2183a413966

Subscribe to this mod's changes

deepseek-harness is a skill published in the GitHub repository Seann0824/deepseek-harness-for-codex (12 stars, last pushed 2mo ago), licensed MIT. It adds 62 tokens to every session and 629 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Browser automation Building agents

deepseek-harness next to 6 skills from other repositories
Mod	Type	Stars	Last push	Licence	Scan	Tokens	Works with
deepseek-harnessSeann0824/deepseek-harness-for-codex	Skill	12	2mo ago	MIT	A	62	Claude Code, Codex
interactive-loginyc-software/qm	Skill	16k	yesterday	MIT	A	46	Claude Code, Codex
pinchtab-mcppinchtab/pinchtab	Skill	11k	10d ago	MIT	A	52	Claude Code, Codex
azure-messaging-webpubsub-javamicrosoft/skills	Skill	3.2k	yesterday	MIT	A	43	Claude Code
cua-driverdavidondrej/skills	Skill	4.3k	3d ago	MIT	A	68	Claude Code, Codex
open-design-browsercompozy/compozy	Skill	2.8k	yesterday	MIT	A	56	Claude Code, Codex
southwestborski/travel-hacking-toolkit	Skill	895	yesterday	MIT	A	40	Claude Code