generated: '2026-09-12'
method: searched
source: >-
  https://project-hami.io/.well-known/agent-skills/index.json (Agent Skills discovery 0.2.0) and the
  skill/ directory of https://github.com/Project-HAMi/HAMi
note: >-
  Every skill here is PROVIDER-AUTHORED and saved verbatim — nothing in this directory was generated by
  API Evangelist. HAMi publishes an Agent Skills discovery index at the well-known path on its docs site
  and ships two further Agent Skills in the skill/ directory of the main repository. A third repository
  entry, skill/hami-gpu-debug/SKILL.md, is present but is a zero-byte placeholder and is recorded below
  rather than saved.
discovery:
  index: https://project-hami.io/.well-known/agent-skills/index.json
  schema: https://schemas.agentskills.io/discovery/0.2.0/schema.json
  http_status: 200
  fetched: '2026-09-12'
skills:
  - name: hami-docs
    file: hami-docs.md
    source: https://project-hami.io/.well-known/agent-skills/hami-docs/SKILL.md
    http_status: 200
    content_type: text/markdown
    digest: sha256:412deb896b3f005e65c4aafeb8c2d37dbb7c188b4100f4bb3020b53905fec7a3
    description: Discover, search, and cite HAMi documentation and release resources.
    discoverable: true
    method: searched
  - name: hami_vgpu_metrics_summarizer
    file: hami-vgpu-metrics-summary.md
    source: https://github.com/Project-HAMi/HAMi/blob/master/skill/hami-vgpu-metrics-summary/SKILL.md
    http_status: 200
    description: >-
      Summarize HAMi vGPU metrics from Prometheus-style /metrics output — organizes GPU allocation by
      node, device, pod and namespace and reports core allocation, memory allocation, allocation-based
      utilization, sharing density and namespace usage patterns.
    discoverable: false
    note: Shipped in the repository only; not listed in the well-known discovery index.
    method: searched
  - name: k8s-debug-pending-pod
    file: hami-k8s-debug-gpu-pod.md
    source: https://github.com/Project-HAMi/HAMi/blob/master/skill/k8s-debug-gpu-pod/SKILL.md
    http_status: 200
    description: >-
      Event-driven triage for pods stuck in Pending, CrashLoopBackOff or ImagePullBackOff, then a deep
      dive into scheduling failures, resource exhaustion, image-pull errors and crash loops.
    discoverable: false
    note: Shipped in the repository only; not listed in the well-known discovery index.
    method: searched
not_saved:
  - name: hami-gpu-debug
    source: https://github.com/Project-HAMi/HAMi/blob/master/skill/hami-gpu-debug/SKILL.md
    http_status: 200
    reason: Zero-byte file in the repository tree — an empty placeholder, not a skill.