Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/444136347/dsh-capability-discoverynpx agentmods add skills/444136347/dsh-capability-discovery/capability-discoveryWrote 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/444136347/dsh-capability-discovery/capability-discovery)<a href="https://agentmods.dev/skills/444136347/dsh-capability-discovery/capability-discovery"><img src="https://agentmods.dev/badge/skills/444136347/dsh-capability-discovery/capability-discovery.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 | $0.00039 | $0.00912 |
| Opus 5 | $0.00019 | $0.00456 |
| Sonnet 5 | $0.00008 | $0.00182 |
| Haiku 4.5 | $0.00004 | $0.00091 |
Grade A, and why
capability-discovery 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 7d 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 — 64 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Capability discovery for DeepSeek Harness
Use this skill when the user is looking for an existing capability rather than asking you to build one from scratch.
1. Search before recommending
Extract one to three concrete English keywords from the request. Run one combined search:
node ../../cli/dsh-capability.mjs search <keywords...> --limit 8 --json
If the user explicitly asks for a Skill, add --type skill. MCP Server searches use the strict --type mcp filter, which returns only candidates with explicit server evidence. If the user asks for a management console, client, bridge, or other MCP-related integration, do not use --type mcp; search the broader capability set and explain the returned type and mcpRole.
Do not repeat the same search command. Retry once only if the first search returns no relevant candidates, using a different or broader query. Do not rerun a successful search merely to reformat or summarize its JSON.
Read sourceErrors: a partial multi-source result is usable, but disclose failed sources instead of presenting it as complete coverage. Do not retry only to make every source succeed.
2. Prefer evidence over popularity
Use the returned score as a starting point, then check the top candidates for purpose fit. sources shows how many independent indexes surfaced the same repository. Explain that GitHub ★ means GitHub Stars, not a rating. Stars are a weak signal, not proof of quality or safety.
Return at most three candidates unless the user asks for a longer list. Explain the practical difference between them.
3. Keep the answer compact
Use this exact compact order unless the user requests another format:
- One sentence stating how many candidates were found and which one is recommended.
- One Markdown table using the user's language. For Chinese, use
候选 | 证据 | 适用场景and format evidence like插件 · 相关度 66.9 · GitHub ★1. Otherwise useCandidate | Evidence | Best forand format evidence likeplugin · relevance 66.9 · GitHub ★1. Display only the repository name without its owner as a link to the full URL. - One short recommendation paragraph explaining the main tradeoff.
- One source-status line: report
successful/total. Do not list successful source names; name failed sources and short reasons only when failures exist. - Always end with the operation status. Use the user's language for labels and operation status, such as
未安装任何内容。for a Chinese request.
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.
- 7d ago First seen · 64 lines · 39 tokens per session scan A 17981748f681
capability-discovery is a skill published in the GitHub repository 444136347/dsh-capability-discovery (3 stars, last pushed 19d ago), licensed Apache-2.0. It adds 39 tokens to every session and 912 once invoked, about $0.0002 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 2026-08-31.
Other skills, from other repositories
clarify-intent-and-establish-shared-understanding
Grounded in first principles, rigorously examine and refine a user's plan, task, decision, goal, strategy, proposal, or idea through structured, progressively deeper questioning, in order to bridge the gap between the User and the Agent. Use when the user explicitly requests grilling, challenge, pressure-testing…
dsh-web-skin-developer
Build a new skin for the dsh-web skin collection (DSH Web GUI) and publish it into the Skin Center — the first-level settings section — scaffold with scripts/dsh-skin-new, author the v2 skin.json manifest plus skin.css token remap (pure asset directory, no package.json, no build step), validate with scripts/dsh-skin…
dsh-sdk-upgrade
Safely select and install a compatible official @deepseek-ai SDK release for dsh plugin projects (dsh-web, dsh-trading, and similar monorepos) from npm using an isolated worktree, explicit cohort review, CI-equivalent validation, and controlled rollout — including syncing the project's declared DSH host-version floor…
dsh-web-sdk-compatibility
Adapt and repair dsh-web after an approved official @deepseek-ai SDK/runtime cohort is selected or installed. Compare public API, type, service-injection, module-table, protocol, and behavior changes; map every change to repository consumers; implement the smallest fixes and durable compatibility contracts; handle…
manage-taskboard
Manage work in the native DeepSeek Harness Taskboard with exact task ids and optimistic versions. Use when an Agent must inspect project work, claim an eligible todo, record progress or blockers, verify an implementation, submit it for human review, or release its own claim; also use when a human asks how to accept…
yao-ocr
OCR text recognition expert. ALWAYS invoke this skill when you need to extract text from images or PDFs — including invoices, receipts, ID cards, bank cards, business licenses, tables, handwritten documents, or any visual text content.