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 Orkas-AI/Orkas-VideoStudio --skill stage-composegit clone --depth 1 https://github.com/Orkas-AI/Orkas-VideoStudioWrote 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/orkas-ai/orkas-videostudio/stage-compose)<a href="https://agentmods.dev/skills/orkas-ai/orkas-videostudio/stage-compose"><img src="https://agentmods.dev/badge/skills/orkas-ai/orkas-videostudio/stage-compose/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.
<a href="https://agentmods.dev/skills/orkas-ai/orkas-videostudio/stage-compose"><img src="https://agentmods.dev/badge/skills/orkas-ai/orkas-videostudio/stage-compose.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00096 | $0.03311 |
| Opus 5 | $0.00048 | $0.01656 |
| Sonnet 5 | $0.00019 | $0.00662 |
| Haiku 4.5 | $0.00010 | $0.00331 |
Grade A, and why
stage-compose 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 yesterday.
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 — 112 lines — stays where its author put it; the contents beside it link to each section on GitHub.
stage-compose
How to author an Orkas/OVS HTML composition and turn it into a video. composition-manifest.json v2 is the canonical timeline/audio/design artifact. In this open-source build, ovs draft is the VideoStudio-style production gate: it runs manifest/source/narration/local-asset checks, HyperFrames check/render, media QA, sampled-frame video QA, and writes one report. HyperFrames remains the render backend; do not bypass the draft gate for user-facing drafts.
For visual direction, apply frontend-design before writing manifest art_direction. If the user provides a reference image/video, DESIGN.md, brand guide, screenshot, design notes, existing app UI, or named style, apply design-system-importer to convert it into intent-bound reference constraints and compact tokens. composition-design-review is a bounded full-frame review before a visual preview is shown, with a post-draft fallback when preview is skipped.
Gate authorization belongs to gate-control. This skill supplies the canonical manifest, contact sheet, draft, and QA evidence; after any Gate B/Preview/Gate D decision or resumed approval, run ovs gate transition rather than inventing a second confirmation or recovery loop.
When the production runtime is unavailable, return a clearly unexecuted candidate package instead of fabricating a render: locked assumptions, final-form script/narration, timed storyboard, exact visible copy/captions, visual/audio and rights-safe asset plan, target export settings, and checkable preview/final QA.
Fast COMPOSE Runbook
After Gate B approves the canonical composition manifest, keep the production turn narrow:
- Read the approved
project/composition/composition-manifest.jsonand this skill if not already loaded for the current turn. Also readfrontend-design; readdesign-system-importeronly when a concrete style source or explicit named reference exists. Readcomposition-design-reviewafter snapshot evidence exists, or after the draft when preview was intentionally skipped and the fallback trigger applies. - If standalone narration is needed, capture the exact
ovs speech-capabilitiesprofile in the approved plan and runovs narration fitbefore synthesis. Visual authoring,ovs check, andovs snapshotmay proceed while narration is pending so the visual candidate remains reviewable. Required narration blocks only completeovs draft/final delivery. Before drafting, runovs speakonce toproject/composition/assets/narration.mp3, declare the composition-owned track, probe its duration, rerun measured fit, and reconcile. Reuse an existing matching narration file; do not synthesize again just because a visual preview changed. For an AUTO segment, useaudio.owner: "assembler"and render silent. - Keep the approved
project/composition/composition-manifest.jsonwithschema_version: 2, including approved timeline/copy/source mappings, audio ownership, andart_direction. Confirm its dedicatedcover, anyreferences/reference_fidelity, andVisualDirectionV1; then prepare and author static resolved/hero frames before deterministic motion. The exact 0s cover must already be readable. Reconcile after manifest timing/audio changes without replacing authored DOM/CSS/SVG/motion. - Decide whether to open the optional HTML Preview Gate before rendering mp4. Use the preview gate when expected render rework is expensive: target duration >= 20s, scene count >= 3, render cost is likely slow, or the composition has dense text, complex SVG/GSAP, many branded/supplied assets, tight narration timing, or a prior draft failure. Skip it for short/simple work: target duration < 20s, scene count <= 2, no narration/timing complexity, and no obvious visual-risk signal. The subject category alone never forces the preview gate.
- If visual preview is needed, run
ovs checkandovs snapshot. Review the complete contact-sheet index and open the cover, QA-named frames and risky cells at full size; compare declared references. Collect all blockers, apply one localized repair, then rerun check/snapshot and review the complete new revision. The contact sheet is only an index. The author checklist is advisory and adds no scored approval gate; present current passing snapshot evidence after addressing concrete visible blockers. - Run the draft command:
ovs draft project/composition --out project/render/draft.mp4 --quality draft --report project/render/draft-report.json --findings project/composition/qa/check.json. Before rendering, this gate validates manifest/HTML consistency, prepares declared local vendor assets, blocks remote runtime resources, verifies local assets, checks shotlist/source alignment, and checks narration mapping. Then it runs HyperFrames check/render, media QA, sampled-frame QA, and writes one report. - If draft fails, repair the highest canonical source, reconcile when needed, and retry only after the authored signature changes. Do not delete QA state or repeat an unchanged strategy. After two non-converging passes, preserve the current artifact and show the visible unresolved issue with concrete options through gate-control; await a real user direction before another cycle.
- If the draft command returns
ok: true, the composition is frozen for final-video confirmation. Do not editindex.html,composition-manifest.json, assets, or narration again in the same turn unless the report contains a real blocker (check/source/audio/video QA failure), or the user explicitly asks for a revision. Visual/readability warnings and design-reviewfix/polishnotes are advisories, not permission to self-repair. - If preview was skipped and the fallback trigger applies, run
composition-design-reviewagainst representative draft frames. Open final-video confirmation only after the draft command returnsok: trueand any triggered review has no concrete blockers.
What ships with it
4 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.
- yesterday Changed · -128 lines 7f1e25240625
- 12d ago First seen · 240 lines · 96 tokens per session scan A 1293fd7e0070
stage-compose is a skill published in the GitHub repository Orkas-AI/Orkas-VideoStudio (486 stars, last pushed 2d ago), licensed MIT. It adds 96 tokens to every session and 3,311 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 2026-08-30.
Other skills, from other repositories
claude-real-video
Watch a video for the user. Use when the user shares a video URL (YouTube etc.) or local video file and wants it summarized, analyzed, or discussed — Claude can't ingest video directly, so this skill extracts scene-aware keyframes + transcript first, then reads those.
brain-ingest
The process for digesting a conversation, document, or research result, classifying it, and writing it down as brain content (a root-page update or a new/updated page) through the brain CLI.
issue-resolve
Use when an assessed gflow-cli issue (verdict CONFIRMED-BUG or LIKELY-BUG) has localized, verifiable scope and should be driven to a fix. Mutating and gated: it works in an isolated worktree, fixes test-first, and opens a DRAFT PR for human review. Built to run autonomously (hermes-ops) within a strict action envelope…
issue-assessment
Use when triaging a GitHub issue for gflow-cli — a reporter's bug claim, a freshly-filed issue, or deciding whether and how to act on one. Also use when an autonomous agent (hermes-ops) picks up a labelled issue. Read-only: produces a verdict, an end-to-end-verifiability judgment, and a reporter-facing reply. Does not…
live-verify
Two-part gate for gflow-cli feature/fix work. Part 1 (Pre-flight): use when starting work on a gflow-cli feature or fix — confirms the checkout reflects current develop before investing effort. Part 2 (Live-verify): use before claiming gflow-cli work done, especially anything touching a generation code path…
sonar
Check the SonarCloud quality gate for a PR (or the current branch) and drive it to zero. Reports GREEN, or the exact PR-scoped new issues/hotspots/coverage gaps to fix.