Execute qualquer Skill no Manus
com um clique

Execute qualquer Skill no Manus com um clique

ai-run-manifest

Estrelas2

Forks1

Atualizado22 de maio de 2026 às 03:02

Add or audit replay-manifest sidecars for AI scripts that create datasets, feature tables, model artifacts, evaluation reports, or fetch/download roots.

Instalação

Instalar com Codex ou Claude Copie este prompt, cole no Codex, Claude ou outro assistente e deixe que ele revise a página da skill e instale para você.

Executar no Manus

Fonte

VMAFx

VMAFx/vmafx

Abrir repositório GitHub Ver repositórios do creator

Download

Executar no Manus

Ocupações relacionadasSOC

Baseado na classificação ocupacional SOC

Desenvolvedores de softwareInformática e Matemática·SOC 15-1252

SKILL.md

readonly

name	ai-run-manifest
description	Add or audit replay-manifest sidecars for AI scripts that create datasets, feature tables, model artifacts, evaluation reports, or fetch/download roots.

/ai-run-manifest

When to use

A script under ai/scripts/ writes a durable artifact: parquet, JSONL, ONNX, registry update, model-card sidecar, evaluation report, calibration JSON, or downloaded dataset/cache root.
A PR changes the inputs, filtering, schema, or defaults for one of those artifact-producing scripts.
You are auditing stale AI artifacts and need to prove which command produced each local file.

Required Pattern

For scripts executable as python ai/scripts/<name>.py, call bootstrap_ai_script(__file__) from ai/scripts/_script_bootstrap.py before importing aiutils or sibling ai/scripts modules. Add only the optional roots the script actually needs (include_repo_root, include_ai_scripts, include_vmaf_tune_src).
For a new standalone sidecar, call aiutils.run_manifest.write_run_manifest(). Pass:
- schema: script-specific, versioned, kebab-case plus -v1.
- entrypoint: the current script path.
- repo_root: the repository root.
- argv: the raw CLI arguments.
- args: parsed CLI namespace or mapping.
- inputs / outputs: named paths, even when optional/missing.
- sections: adapter-specific counters, selected features, gates, config, schemas, or status.
If the script already has a stable report schema, embed only build_run_provenance() in that existing report instead of renaming the report to a new manifest schema.
Default new sidecars beside the artifact (<output>.manifest.json) unless the artifact has a stronger local convention.
For a batch manifest runner, import make_argument_parser(), collect_cli_argv(), and add_batch_manifest_arguments() from aiutils.cli_helpers instead of duplicating the common --manifest / --base-dir / report-output / fail-fast flags.
Add tests that assert the script writes the sidecar and that run_provenance.schema == "ai-run-provenance-v1".

Guardrails

Do not hand-roll path hashing, JSON sorting, or run_provenance structure in the script.
Do not add new ad hoc sys.path.insert(...) blocks to AI scripts; extend _script_bootstrap.py when a new repo-local import root is genuinely needed.
Keep row schemas stable. Put run-level evidence in the sidecar, not every row.
Preserve existing machine-readable report keys unless the PR explicitly bumps that report schema.
Fetchers record URLs, archives, selected rows, and cache roots. Feature-table builders record selected columns, row/source counts, and source artifacts. Model exporters record input checkpoint/config, ONNX outputs, and gate status.
Update ai/AGENTS.md, the relevant docs under docs/ai/, and the model card or runbook that tells humans how to reproduce the artifact.

Mais deste repositório

mesmo repositório

sync-upstream

VMAFx/vmafx

Reconcile fork master with Netflix/vmaf master. Detects the fork's port-only topology (no shared merge-base) and emits a coverage report; falls back to merge-based sync when the histories are actually connected.

2026-06-272

add-feature-extractor

VMAFx/vmafx

Scaffold a new feature extractor (e.g. a novel metric) with C source+header, registry entry, doc stub, and a smoke test. Does not produce a SIMD or GPU path — those come via /add-simd-path and /add-gpu-backend.

2026-06-032

add-gpu-backend

VMAFx/vmafx

Scaffold a complete new GPU backend (hip, rocm, metal, opencl, etc.) with runtime, feature kernel stubs, public header, Meson options, CI workflow, doc stub, and smoke test. Flagship scaffolding skill. Established precedents on the fork — CUDA (legacy), SYCL (legacy), Vulkan (T5-1, ADR-0175 scaffold + ADR-0186 image-import).

2026-06-032

add-simd-path

VMAFx/vmafx

Scaffold a new SIMD implementation for an existing feature. Creates intrinsics source + header + a bit-exact-vs-scalar comparison test; wires into runtime dispatch. Supports kernel-spec flags that pull recurring patterns from simd_dx.h.

2026-06-032

add-k8s-resource

VMAFx/vmafx

Scaffold a new Kubernetes CRD + kubebuilder controller + RBAC + helm chart entry for the vmafx-operator. Follows the VmafxJob / VmafxNode / VmafxModelTraining precedent in cmd/vmafx-operator/ (ADR-0714, ADR-0709 parent).

2026-05-312

add-mcp-tool

VMAFx/vmafx

Scaffold a new VMAFX MCP tool handler with byte-compatible Go (cmd/vmafx-mcp) and Python (mcp-server/vmaf-mcp) implementations, registration, smoke tests, and docs/mcp/ page. Companion to /add-gpu-backend and /add-feature-extractor for the MCP surface.

2026-05-312

/ai-run-manifest

When to use

A script under ai/scripts/ writes a durable artifact: parquet, JSONL, ONNX, registry update, model-card sidecar, evaluation report, calibration JSON, or downloaded dataset/cache root.

A PR changes the inputs, filtering, schema, or defaults for one of those artifact-producing scripts.

You are auditing stale AI artifacts and need to prove which command produced each local file.

Required Pattern

For scripts executable as python ai/scripts/<name>.py, call bootstrap_ai_script(__file__) from ai/scripts/_script_bootstrap.py before importing aiutils or sibling ai/scripts modules. Add only the optional roots the script actually needs (include_repo_root, include_ai_scripts, include_vmaf_tune_src).

For a new standalone sidecar, call aiutils.run_manifest.write_run_manifest(). Pass:

schema: script-specific, versioned, kebab-case plus -v1.
entrypoint: the current script path.
repo_root: the repository root.
argv: the raw CLI arguments.
args: parsed CLI namespace or mapping.
inputs / outputs: named paths, even when optional/missing.
sections: adapter-specific counters, selected features, gates, config, schemas, or status.

If the script already has a stable report schema, embed only build_run_provenance() in that existing report instead of renaming the report to a new manifest schema.

Default new sidecars beside the artifact (<output>.manifest.json) unless the artifact has a stronger local convention.

For a batch manifest runner, import make_argument_parser(), collect_cli_argv(), and add_batch_manifest_arguments() from aiutils.cli_helpers instead of duplicating the common --manifest / --base-dir / report-output / fail-fast flags.

Add tests that assert the script writes the sidecar and that run_provenance.schema == "ai-run-provenance-v1".

Guardrails

Do not hand-roll path hashing, JSON sorting, or run_provenance structure in the script.

Do not add new ad hoc sys.path.insert(...) blocks to AI scripts; extend _script_bootstrap.py when a new repo-local import root is genuinely needed.

Keep row schemas stable. Put run-level evidence in the sidecar, not every row.

Preserve existing machine-readable report keys unless the PR explicitly bumps that report schema.

Fetchers record URLs, archives, selected rows, and cache roots. Feature-table builders record selected columns, row/source counts, and source artifacts. Model exporters record input checkpoint/config, ONNX outputs, and gate status.

Update ai/AGENTS.md, the relevant docs under docs/ai/, and the model card or runbook that tells humans how to reproduce the artifact.