akari
Create, open, or resume AKARI Video projects. Use when asked to make a video in a folder without `.akari/`, or to check an AKARI project's state.
- License 许可
- MIT
- 最近更新
- 2026-10-22
- Stars
- ★ 214
- Forks
- 25
01 前言 Front Matter
- name
- akari
- description
- Create, open, or resume AKARI Video projects. Use when asked to make a video in a folder without `.akari/`, or to check an AKARI project's state.
02 资源文件 1
- 脚本test/skills-command.test.mjs192 B
03 SKILL.md 正文
Start or resume AKARI Video
Language: Respond in the user's language — conversation, questions, approval prompts, and the reports and records you write (including decision-log rows) follow the language the user writes in.
For project editing, read plan-edit autonomy and gates before proceeding.
The library defaults to the workspace's library/. Without a workspace, use the legacy ~/.akari/assets/.
Check the first line of akari-assets list (or akari assets list) for the actual location.
Below, <ライブラリの置き場> is the existing library-location placeholder; audio belongs in its audio/ subdirectory.
This entry point handles only the following 3 branches. Do not duplicate the main skills' procedures here.
Use the akari CLI on PATH. If unavailable, use
node "${AKARI_HOME:-$HOME/.akari}/app/packages/akari-launcher/bin/akari.mjs";
if that is also unavailable, run packages/akari-launcher/bin/akari.mjs from an actual repository checkout with
node. If none exists, explain what is missing and stop; do not invent paths.
Tool procedures for external tools (akari recipe)
For work using external tools or generative AI, if a matching recipe exists, run akari recipe plan <id>
to check connector candidates and gates. Follow the gate procedure when a gate blocks execution.
Execute with the existing verbs, then run akari project sync and record each output as a manifest
tool-output with recipe: "<id>@<version>", connector, and inputs_sha256.
Recommend a playbook
Recommend at most twice: when starting a project and immediately after the user places new source material.
Read .akari/intake.json and available analysis.json records for target duration, format, speakers, and media kinds.
Run akari playbook list --json (add --channel <name> for the current channel). Compare each entry's
category, tags (especially form, length, and materials), and requires with that evidence.
Rank up to three matches and explain the concrete evidence for each in chat. For example, portrait source
material and a target of 60 seconds or less favor vertical explainers or short dramas; a talking head plus
screen capture favors an explainer; music-led footage favors an MV. A missing required tool lowers the rank
and must be mentioned. Link to the playbook's review sheet path when one exists, or state that
.akari/reports/playbook-apply.html will be generated by applying it. Never apply a recommendation
automatically. After the user chooses, run akari playbook apply <id> --into <project>.
Resume
- Search for
.akari/from cwd up through its ancestors. Call the discovered project<project>. A.akari/root.jsonwithschema: "creator-root/v1"marks a workspace, not a video project. If only a workspace is found, proceed to Create or Open according to the request. - Run
akari status "$PWD" --json. Usenext_skill/waiting_onto explain the next action in 1–2 sentences. Report failure orstate_health: inconclusiveas unavailable state; do not infer the workflow stage from fragments of files. - If
next_skillis present, read directly by path:<project>/.claude/skills/<next_skill>/SKILL.md. If absent, report the missing file rather than inventing a procedure. Ifwaiting_onis present, explain that action without skipping the main skill's wait conditions.
Create
Use when .akari/ is absent or a new project is explicitly requested. The legacy matcher
新規作成を明示された場合 denotes an explicit request for a new project.
Even if no workspace exists, do not create one without the user's consent (propose → consent → create).
The machine-matched cue 利用者の同意なしに作成しない means that creation requires the user's consent.
- Detect a workspace through these 2 routes (either is sufficient).
- Check whether
lastRootin<AKARI_HOME>/creator-root.json(default:~/.akari/creator-root.json) exists. - Search from cwd through its ancestors for
.akari/root.json, and verify thatschemais"creator-root/v1". Read the discovered workspace's marker too. Report damaged JSON or unknown schemas without overwriting them. Before asking where to create the project, infer the content language from the user's conversation and source audio, independently of the UI language. If they appear to disagree, include a content-language choice in the question in step 2 so the user can settle both the destination and language in one reply.
- Check whether
- Ask only 1 question: “Create this in the workspace
~/Akari/, or in this folder? If there is no workspace, I will create it withakari init.” If an existing workspace was found, replace~/Akari/with its actual path. Include the specific destination for the workspace option in the same question:<作業場>/channels/<channel>/videos/<日付-スラッグ>/(the existing workspace/date-slug placeholders). Use the first entry inroot.json'schannelsas<channel>, ormy-channelif absent. Suggest a short slug from the date and request; propose a different name if a project already has that name. If this folder is selected,<target>is cwd. - After consent, run
akari initonly if the workspace was selected and does not yet exist. Its no-argument ensure operation is idempotent: if a workspace exists, stdout's first line returns its path. If that differs from the approved location, do not create somewhere else on your own. - Before creation, use the selected content language; an explicit choice takes precedence over the inference.
Run
akari new "<target>", adding the --lang option with the BCP 47 code if available. Otherwise write the code to intake.json's language field after creation. Let the CLI add only missing files without overwriting existing ones and enforce the F18 guard against git init inside an existing repository. Main skills remain physical copies;AKARI-SKILLS-VERSIONrecords their version at creation. Resolve in order: explicit --lang → intake.json language → analysis.json transcript.language → ja. Use the primary language subtag for pack selection (en-US → en); preserve ja and use en for a missing pack. - Explain the creation report path printed to stdout and that
plan-editis next, then read directly by path:<target>/.claude/skills/plan-edit/SKILL.md. Skills created during a session may not appear in discovery. They will appear when the agent is started in that project next time, so read them directly in this session.
The 1 question above also obtains consent. Do not ask again if the destination has already been explicitly approved.
If the CLI cannot run, manual workspace-only creation is limited to the approved location. Create
akari.md, channels/<channel>/videos/, library/, inbox/,
.akari/memory/, and .akari/cache/.
Use this stub for akari.md:
# akari.md
この作業場(CreatorRoot)の規約・好みを書く場所です。
AKARI Video のエージェントは動画を作る前に、まずこのファイルを読みます。
## 好み
(まだ何も書かれていません)
Always follow these 2 rules for manual creation.
- Never overwrite any existing file. If
root.jsonorakari.mdexists, do not write it; return to detection. The machine-matched cue既存ファイルを一切上書きしないstates this same no-overwrite rule. - Write
root.jsonlast. Place the workspace marker only after all directories andakari.mdare complete. The machine-matched cue`root.json` は最後に書くstates the ordering rule.
Set .akari/root.json to
{"schema":"creator-root/v1","createdAt":<ISO8601>,"channels":[<channel>]}.
Do not write <AKARI_HOME>/creator-root.json in the fallback.
Do not claim project creation is complete without the CLI; report the missing akari new.
Open
Use only the detection procedure from Create to find the workspace. List existing project directories under
<作業場>/channels/*/videos/* (the existing workspace placeholder) and let the user choose.
If none exists, explain that and point to Create.
Set cwd to the selected project and proceed to Resume. Do not create a workspace during a read-only action.
04 同仓库更多 Skills
verify-project
Verify an AKARI Video project with the L0 / L1 / L2 verification ladder required by an AKARI Video task contract on the current Theia stack. Read when acceptance criteria specify levels such as L0 or L0+L1, to determine exactly what to run and how at each level.(旧名: verify)
verify-edit
Verify edit.json and optional analysis.json / captions.json / media with a deterministic CLI, then finish QA with frame inspection and reporting after PASS. Use immediately after writing or modifying edit.json, before export, and after applying review feedback.(旧名: edit-lint)
setup-remote
Set up remote access to approval reports and previews from another device such as a phone, and send footage to the workspace. Use doctor to assess Tailscale, guide installation and login (human actions), expose the preview server (default 4567) and approval report helper over tailnet-only HTTPS with tailscale serve, connect Taildrop reception to workspace inbox/, and verify access from another device. Use for requests to view reports on a phone, approve remotely, send footage from a phone, or set up remote access. Public internet exposure (funnel) is outside the default scope.
setup-project
Set up an AKARI Video project headlessly. Recursively copy `templates/project-default/`, record the template version, initialize git only when safe, and generate an HTML creation report. No app launch is needed. Use to create a video project or fill in an existing folder as an AKARI Video project.(旧名: create-project)
setup-library
Use for first-time AKARI Video setup or when the current project needs more usable assets. This first-run skill checks ffmpeg / whisper-cli (and can install with approval), reads the bundled catalog with akari assets to propose a starter pack, obtains explicit human approval, then acquires, places, validates, and updates INDEX entries. It also covers the audio library (BGM and sound effects): bulk acquisition of first-party AKARI Sounds from GitHub Releases, a free-source candidate HTML list with drop-folder matching or assisted acquisition for missing families, and keep/drop decisions in the listening gallery. Use for requests such as "I want more BGM" or "add sound effects to the library".(旧名: setup-audio-library)
setup-connections
Set up and inspect AKARI Video connections: AI-service keys, which models are allowed, cost-approval policy, the tool and media-service state, and 3D settings. Use for first-time setup, adding a provider or key later, checking what is usable (「何が使える?」 'what can I use?'), setting a key such as fal (「fal の鍵を設定したい」), 3D settings (「3D の設定を使いたい」), connection checks, and gates before paid generation or external publication. Model selection belongs to the production skills (they read the ledger and preferences); this skill decides allowed models and the cost policy. Works with `.akari/connections.json` and a free, read-only doctor.(旧名: manage-connections)(旧名: connect)