Vss Generate Video Calibration
Generate video calibration outputs through the NVIDIA VSS workflow with an agent that follows the published skill execution path.
Overview
Vss-generate-video-calibration is an agent skill for the Build phase that runs NVIDIA VSS-oriented video calibration generation workflows under NVSkills-Eval-tested skill execution.
Install
npx skills add https://github.com/nvidia/skills --skill vss-generate-video-calibrationWhat is this skill?
- NVSkills-Eval external profile benchmarked with 6 evaluation tasks and 2 attempts per task
- Agents exercised: claude-code and codex with security, correctness, discoverability, effectiveness, and efficiency dimen
- Pass threshold documented at 50% for the evaluation run
- Overall publication verdict recorded as FAIL pending remediation before broad workflow reliance
- 6 evaluation tasks in the NVSkills-Eval dataset
- 50% pass threshold
Adoption & trust: 1 installs on skills.sh; 1.1k GitHub stars; trending (+100% hot-view momentum).
What problem does it solve?
You need reproducible video calibration outputs in a GPU video stack but lack a governed agent workflow that was validated for safety and correctness.
Who is it for?
Builders already in NVIDIA VSS or video analytics experiments who want eval-documented agent behavior for calibration generation tasks.
Skip if: General image editing, unrelated web app CRUD, or teams that need a skill with a passing NVSkills-Eval verdict without their own re-validation.
When should I use this skill?
When an agent needs to execute the vss-generate-video-calibration workflow for relevant video calibration tasks in an NVSkills-Eval external local profile context.
What do I get? / Deliverables
The agent loads the VSS calibration skill, follows the evaluated workflow dimensions, and produces calibration-oriented outputs you can inspect before downstream video pipelines consume them.
- Video calibration generation outputs per the skill workflow
- Traceable skill execution consistent with NVSkills-Eval skill_execution checks
Recommended Skills
Journey fit
Video calibration generation is an integration step in the product pipeline, so Build is the canonical shelf where agents wire media/vision tooling. Integrations subphase fits skills that call external NVIDIA evaluation or VSS pipelines rather than shipping end-user UI alone.
How it compares
Specialized NVIDIA video-calibration skill package, not a generic FFmpeg scripting prompt or an MCP media server.
Common Questions / FAQ
Who is vss-generate-video-calibration for?
Developers and ML engineers wiring NVIDIA video perception workflows who delegate calibration generation steps to Claude Code, Codex, or similar agents.
When should I use vss-generate-video-calibration?
During Build integrations when you are producing or refreshing video calibration artifacts as part of a VSS or external-profile evaluation pipeline.
Is vss-generate-video-calibration safe to install?
Check the Security Audits panel on this page; the bundled NVSkills-Eval summary includes explicit security checks for secret leakage and destructive commands, and the documented run reported an overall FAIL verdict you should weigh before trusting unattended execution.
SKILL.md
READMESKILL.md - Vss Generate Video Calibration
# Evaluation Report Evaluation of the `vss-generate-video-calibration` skill before publication through NVSkills-Eval. This benchmark summarizes 3-Tier Evaluation from NVSkills-Eval results for the skill. The goal is to document whether the skill is safe, discoverable, effective, and useful for agents before it is published for broader workflow use. ## Evaluation Summary - Skill: `vss-generate-video-calibration` - Evaluation date: 2026-06-01 - NVSkills-Eval profile: `external` - Environment: `local` - Dataset: 6 evaluation tasks - Attempts per task: 2 - Pass threshold: 50% - Overall verdict: FAIL ## Agents Used - `claude-code` - `codex` ## Metrics Used Reported benchmark dimensions: - Security: checks whether skill-assisted execution avoids unsafe behavior such as secret leakage, destructive commands, or unauthorized access. - Correctness: checks whether the agent follows the expected workflow and produces the correct final output. - Discoverability: checks whether the agent loads the skill when relevant and avoids using it when irrelevant. - Effectiveness: checks whether the agent performs measurably better with the skill than without it. - Efficiency: checks whether the agent uses fewer tokens and avoids redundant work. Underlying evaluation signals used in this run: - `security` (Security): checks for unsafe operations, secret leakage, and unauthorized access. - `skill_execution` (Skill Execution): verifies that the agent loaded the expected skill and workflow. - `skill_efficiency` (Efficiency): checks routing quality, decoy avoidance, and redundant tool usage. - `accuracy` (Accuracy): grades final-answer correctness against the reference answer. - `goal_accuracy` (Goal Accuracy): checks whether the overall user task completed successfully. - `behavior_check` (Behavior Check): verifies expected behavior steps, including safety expectations. - `token_efficiency` (Token Efficiency): compares token usage with and without the skill. ## Test Tasks The benchmark dataset contained 6 evaluation tasks: - Positive tasks: 6 tasks where the skill was expected to activate. - Negative tasks: 0 tasks where no skill was expected. - Unlabeled tasks: 0 tasks where positive/negative intent could not be inferred. Task composition is derived from the evaluation dataset when possible. Entries with `expected_skill` set are treated as positive skill-activation cases, while entries with `expected_skill: null` are treated as negative activation cases. ## Results | Dimension | Num | `claude-code` | `codex` | |---|---:|---:|---:| | Security | 8 | 96% (+12%) | 79% (+12%) | | Correctness | 8 | 87% (+1%) | 82% (+26%) | | Discoverability | 8 | 89% (+9%) | 69% (+7%) | | Effectiveness | 8 | 57% (-3%) | 55% (+24%) | | Efficiency | 8 | 71% (+14%) | 53% (+6%) | Score values show skill-assisted performance. Values in parentheses show uplift versus the no-skill baseline when baseline data is available. ## Tier 1: Static Validation Summary Tier 1 validation passed with observations. NVSkills-Eval ran 9 checks and found 4 total findings. Top findings: - MEDIUM QUALITY/quality_correctness: SKILL_SPEC recommended field missing: 'metadata.author' (`skills/vss-generate-video-calibration/SKILL.md`) - MEDIUM SCHEMA/author_missing: Author not specified in metadata (`skills/vss-generate-video-calibration/SKILL.md`) - MEDIUM SECURITY/Unknown (SDI-2): The script uses a curl-pipe-sh pattern to download and execute the `uv` installer from astral.sh without any integrity v (`references/sample-dataset.md:132`) - MEDIUM SECURITY/Unknown (SQP-2): SSL verification is explicitly disabled (`ssl_verify: false`) in the RTSP capture request, and the Python script also im (`references/rtsp.md:106`) ## Tier 2: Deduplication Summary Tier 2 validation reported findings. NVSkills-Eval ran 2 checks and found 4 total findings. Top findings: - HIGH DUPLICATE/duplicate: Duplicate content found across references/sample-dataset.md and references/videos.md: "# iterating