agentsclimarketplace

Workload shape audit

Skill Touchdown-Labs/inference-optimization-agent-pack/skills/workload-shape-audit

Loadable systems-thinking skill pack for full-stack inference optimization.

Install
npx -y skills add Touchdown-Labs/inference-optimization-agent-pack --skill workload-shape-audit

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Classify AI workloads across coding, chat/RAG, voice, media, batch, API, self-hosted, and hybrid paths, then produce a workload profile.

SKILL.md

1.6 KB, 332 tokens by cl100k_base, as published. Nobody here has run it

Workload Shape Audit

Use this when the workload is not clearly understood.

The goal is not to optimize yet. The goal is to name the shape of the work so the next optimization is not random.

Classify The Workload

Pick one primary type and any secondary types:

  • Coding agent.
  • Chat/RAG agent.
  • Voice agent.
  • Diffusion/media agent.
  • Batch/offline agent.
  • Self-hosted serving workload.
  • API-only workload.
  • Hybrid workload.

Questions To Answer

  • What is the user trying to accomplish?
  • What counts as success?
  • What does the system load before the model call?
  • What tools can the agent use?
  • What retrieval is sent?
  • What gets cached?
  • What causes retries?
  • What work happens outside the model?
  • Where does human review happen?
  • What does the bill actually measure?

Output

Fill templates/WORKLOAD_PROFILE.md.

Keep it short. The profile should be loadable in a future agent run without wasting context.

First-Pass Leak Labels

Use these labels:

  • prompt/context bloat;
  • premium model overuse;
  • duplicate retrieval;
  • tool-loop waste;
  • cache miss;
  • retry loop;
  • provider/runtime limit;
  • hardware placement;
  • quality rejection;
  • human rework;
  • unknown.

Stop Condition

Stop when the next skill is obvious:

  • API spend or agent waste: route to api-spend-recovery.
  • Self-host, hybrid, or engine placement: route to runtime-routing.
  • Claimed results or before/after data: route to evidence-review.

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Keep looking

Skills are one crate of 326,984. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.