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.
npx skills add n-n-code/n-n-code-skills --skill deepseek-harnessgit clone --depth 1 https://github.com/n-n-code/n-n-code-skillsWrote 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.
[](https://agentmods.dev/skills/n-n-code/n-n-code-skills/deepseek-harness)<a href="https://agentmods.dev/skills/n-n-code/n-n-code-skills/deepseek-harness"><img src="https://agentmods.dev/badge/skills/n-n-code/n-n-code-skills/deepseek-harness.svg" alt="Measured on agentmods" height="20"></a>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 | $0.00099 | $0.01335 |
| Opus 5 | $0.00049 | $0.00668 |
| Sonnet 5 | $0.00020 | $0.00267 |
| Haiku 4.5 | $0.00010 | $0.00134 |
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 2d 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.
How it starts
The opening of the file, as written. The whole thing — 117 lines — stays where its author put it; the contents beside it link to each section on GitHub.
DeepSeek Harness
Work against the exact Harness version the user owns. The bundled references were reviewed against a pre-stable release whose commands, package contracts, and persisted formats can change incompatibly; runtime help and a matching checkout or release are more authoritative than remembered examples.
Select the lane
- Operate or configure: install or launch Harness, choose a profile, inspect patches, configure providers, manage plugins, or troubleshoot runtime state. Load operation and configuration.
- Automate: drive headless, SDK, or ACP surfaces from another process. Load operation and configuration, then add the implementation skill for the caller language when code changes.
- Extend: author an out-of-tree Cordis plugin, tool, model adapter, capability, settings surface, or installable Bundle. Load extension and development.
- Maintain a source checkout: change Harness packages, applications,
tests, or documentation only when the user explicitly targets that checkout.
Load extension and development
and follow the checkout's complete
AGENTS.mdhierarchy and exposed project skills.
Generic DeepSeek API calls do not select this skill. Use prompt-engineering
as primary when prompt behavior is the deliverable, security when the user
requests a security assessment, project-platform-diagnose while a
platform-dependent failure remains unexplained, and project-release-maintainer
when publishing or release machinery is the main job.
Ground the target
- Identify whether the target is an installed CLI, source checkout, out-of-tree plugin project, or SDK/ACP caller.
- Resolve the exact installed or checkout version before choosing commands,
config fields, imports, or package examples. Use profile-free version help
when available; for source work, inspect
package.json, the applicable instructions, and version-matched documentation. - Prefer target-local source, runtime help, generated catalogs, and package READMEs over the snapshot bundled with this skill. If the executable or matching sources are unavailable, report the gap instead of installing or upgrading them without authority.
- Name the Harness home, profile, and workspace explicitly before any action that can create state, load code, call a model, or change files. Distinguish persistent Harness state from the workspace the agent may inspect or edit.
What ships with it
3 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 2d ago First seen · 117 lines · 99 tokens per session scan A 2039e4661d2b
deepseek-harness is a skill published in the GitHub repository n-n-code/n-n-code-skills (4 stars, last pushed yesterday), licensed MIT. It adds 99 tokens to every session and 1,335 once invoked, about $0.0005 per session on Opus 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 2026-09-05.
Other skills, from other repositories
audit-pr
Audit a whole PR against the delivery contract and return MERGE-READY or evidenced blockers with the full URL. Consumes the current review-change REVIEW-PASS receipt instead of re-running review axes; posts a SHA-bound ready comment; never edits or merges. Triggers: "audit-pr", "is this PR ready", "merge gate".
design-feature
Turn a raw idea or existing feature into a designed product SPEC by completing entity, integration, role, and expectation closure. Upserts never destroy recorded decisions. Triggers: "design-feature", "design this feature", "define product scope".
plan-feature
Route designed features or issues into engineering planning and roadmap registration; undesigned work stops at design-feature. Supports --next, --from-issue, and --scaffold. Triggers: "plan-feature", "plan a feature", "plan the next roadmap feature", "create SPEC and TASKS".
product-audit
Audit the whole product across code, quality, process, docs, roadmap, and tooling. Persist one severity-ranked, F-numbered report with proposals; never fix or file work. Triggers: "product-audit", "audit the product", "full health check", "are we product-ready", "CTO review".
audit-docs
Audit cross-document coherence: docs ↔ roadmap ↔ code ↔ fix index ↔ issues. Finds drift — features in docs/ not in the roadmap (or vice versa), fix-index entries already merged/closed, broken documentation-map links, dependency cycles, artifacts in the wrong language, naming-convention violations — and reports them…
init-workspace
Adapt the workflow scaffold to a new repository or add only missing substrate blocks to an existing install. Every install, hook, and overwrite needs explicit consent. Triggers: "init-workspace", "set up agentic workflow", "upgrade workflow scaffold".