AGENT SKILL

omni-suggest

Suggest marketplace skills that fit this project's detected stack, add the chosen ones through Omni with consent, then offer to assign them to the project. Use when the user asks what skills would help this project, what to install, or how to improve the agent setup.

License 许可
MIT
所属仓库
suarja/omni
最近更新
2026-10-22
Stars
★ 0
Forks
0

01 前言 Front Matter

name
omni-suggest
description
Suggest marketplace skills that fit this project's detected stack, add the chosen ones through Omni with consent, then offer to assign them to the project. Use when the user asks what skills would help this project, what to install, or how to improve the agent setup.

03 SKILL.md 正文

Suggesting skills for this project's stack

Omni already ranks the catalogue deterministically for this project — tag match, installs, audit and already-in-marketplace bonuses, no model involved. This skill reads that ranking, does the one judgment step (which of the top matches are actually worth adding), and routes every write through the existing preview → consent → apply path.

Steps

  1. omni_stack_profile { projectId } — the detected tags with their evidence. If the project carries none, say so; there is nothing to suggest yet.
  2. omni_suggest { projectId } — the ranked candidates, each with why (the matched tags) and alreadyInProject. Present the top 5 that are not already in the project, one line each: name and its why.
  3. Ask which of these the user wants added. No or unclear means stop here.
  4. For each chosen skill: omni_catalog_search { query: name } to find its entry (owner/repo, url). Fetch the SKILL.md at that url yourself — Omni's tools never fetch on the agent's behalf. Use omni-marketplace-routing to pick the plugin it belongs in.
  5. omni_preview_create_skill { sourceId, packageId, folderName, files } with the fetched content, present the plan, then use the consent rule below and call omni_apply.
  6. After the receipt, propose omni_preview_assign { artifactPath, sourceId, packageId, projectIds: [projectId] } for this project — a second, separately previewed gesture; use the consent rule again before calling omni_apply.

Rules

  • Never invent a suggestion outside omni_suggest's list; the ranking is the evidence.
  • Skip anything alreadyInProject reports as already there.

Use the packageId returned by omni_list_marketplace or omni_find_skill, for example packages/<name>. The short package name also works. Previews and receipts keep the canonical observed id.

Consent

<!-- Generated from packages/omni-skills/consent.md. -->

For ordinary skill changes, previously given clear approval for this exact plan counts; do not ask again. If approval is not already clear, present the plan and wait for a clear yes; no or unclear means do nothing. For approval: "omni": Approval happens in Omni for the reason returned with the preview. Tell the user in one line, then call omni_apply; it returns pending while the request waits. Call it again with the same requestId; the request does not expire. If the user approves in Omni later, omni_receipt shows the result. Route every write through Omni preview, omni_apply, and its receipt.

For sharing or publishing, follow omni-share; author and fix skills only in the active workshop, never in a storefront.

Install external skills through Omni using omni_list_external_library, omni_preview_install_external, omni_apply and its receipt; never run npx skills add directly.

Managed files

Never edit a skill folder, a plugin or marketplace manifest, or the personal marketplace directly — Omni's guard refuses it. Use omni_preview_update_skill, omni_preview_capture, or omni_preview_assign to plan the change, then omni_apply using the user's existing clear authorization for that exact change.

04 同仓库更多 Skills