gemini-omni

Generate and conversationally edit short videos with Google Gemini Omni Flash (`gemini-omni-flash-preview`). Use when: (1) iterating on a clip with natural-language edits instead of regenerating ("make the phone invisible, keep everything else the same"), (2) generating 3-10s 720p clips with synthesized audio, rendered on-screen text, or timecoded beats, (3) binding reference images to roles with <FIRST_FRAME>/<IMAGE_REF_N> prompt tags, (4) editing an existing uploaded video. Accessed via the `gemini_omni_video` tool using the project's GEMINI_API_KEY/GOOGLE_API_KEY — the same key as Imagen and Google TTS.

Quality

78%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./.agents/skills/gemini-omni/SKILL.md

Quality

Content

67%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A detailed, actionable guide with a strong numbered editing workflow and concrete parameters. Its main weakness is conciseness: time-sensitive dates and pricing sit in the main intro rather than a versioned/deprecated section, and no full tool-call example is shown.

Suggestions

Move time-sensitive specifics (I/O 2026 announcement date, 'June 30, 2026' access date, exact token/sec pricing) out of the intro into a clearly labeled 'Preview / version notes' or 'Deprecated' section so stale details don't clutter the main flow.

Add one complete example tool call to gemini_omni_video showing the generation→edit handoff (operation, previous_interaction_id, reference_image_paths) so the workflow is copy-paste executable end-to-end.

Add an explicit verification checkpoint to the editing workflow (e.g., 'Confirm the result returns a new interaction_id before refining further') to close the validation gap.

Dimension	Reasoning	Score
Conciseness	Mostly efficient API-specific guidance, but the intro embeds time-sensitive specifics ('announced at I/O 2026', 'developer access since June 30, 2026', exact pricing '5,792 output tokens/sec at $17.50/1M') in the main flow rather than a deprecated/versioned section, which the rubric penalizes.	3 / 5
Actionability	Concrete, actionable guidance with real parameter names (reference_image_paths, previous_interaction_id, operation="edit_video", store, input_video_path), timecode syntax, and copy-paste good/bad edit-prompt pairs; minor gaps as there is no full end-to-end tool-call example.	4 / 5
Workflow Clarity	The conversational-editing workflow is a clearly numbered 4-step sequence with a layer-refinement feedback loop and the 'store' gotcha as a checkpoint, but there is no explicit verify-the-edit-succeeded step.	4 / 5
Progressive Disclosure	Well-organized into clear sections with a Sources list of one-level-deep external doc links and no nested references, but all content lives inline in a single ~120-line file with no bundle split, leaving minor organization gaps.	4 / 5
	Total	15 / 20 Passed

Description

88%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A strong, specific description with clear what/when structure and concrete numbered triggers. It distinguishes the skill via its conversational-editing differentiator, with only minor overlap risk against other video-generation providers.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions — 'Generate and conversationally edit short videos', 'iterating on a clip with natural-language edits', 'binding reference images to roles with <FIRST_FRAME>/<IMAGE_REF_N> prompt tags', 'editing an existing uploaded video' — giving comprehensive coverage of the skill's capabilities.	5 / 5
Completeness	Explicitly answers both 'what' ('Generate and conversationally edit short videos with Google Gemini Omni Flash') and 'when' via a four-item 'Use when:' list with concrete trigger phrases, matching the top anchor.	5 / 5
Trigger Term Quality	Good natural-term coverage (video, clip, edit, regenerate, generate, uploaded video) that users would actually say, but a few synonyms (footage, movie, render) are missing, keeping it just below comprehensive.	4 / 5
Distinctiveness Conflict Risk	The stateful conversational-editing differentiator is a clear niche, but 'generating 3-10s 720p clips' overlaps with sibling video providers (Seedance, Veo, Runway), giving minor overlap risk rather than minimal.	4 / 5
	Total	18 / 20 Passed

Validation

81%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 13 / 16 Passed

Validation for skill structure

Criteria	Description	Result
allowed_tools_field	'allowed-tools' contains unusual tool name(s)	Warning
metadata_version	'metadata.version' is missing	Warning
metadata_field	'metadata' should map string keys to string values	Warning

	Total	13 / 16 Passed

Repository: calesthio/OpenMontage
Path: .agents/skills/gemini-omni/SKILL.md
Commit: 08e2151

Reviewed: 2 days ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.