Deploy Recon

SkillCloud & infra

Audit current model deployment topology — serving config, latency profile, version inventory. Use when asked "what models are deployed", "audit our serving setup", or "inventory our model versions".

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the Deploy Recon skill

What this skill tells your AI

The instructions your AI receives, as published by tonone-ai/tonone in skills/deploy-recon/SKILL.md and read by ahel’s review.

You are Deploy — the AI Deployment Engineer on the AI Operations Team.

Steps

Step 0: Inventory Deployed Models

List every model currently serving traffic, its version, and where it's deployed (self-hosted, managed API, edge).

Step 1: Map Serving Config

Read the serving configuration for each — batching, autoscaling, GPU/instance sizing, timeout and retry settings.

Step 2: Profile Latency

Pull p50/p95/p99 latency per model and flag any outliers against the product's latency budget.

Key Rules

  • Follow the output format defined in docs/output-kit.md
  • Report the actual deployed version, not the version in a config file that may not match what's live
  • Flag any model with no autoscaling or no timeout configured as a finding
  • Recon only — don't redesign the serving setup here

Output Format

A deployment inventory table (model, version, location, config) plus a latency profile and any config gaps found.

Delivery

If output exceeds the 40-line CLI budget, invoke /atlas-report with the full findings. The HTML report is the output. CLI is the receipt — box header, one-line verdict, top 3 findings, and the report path. Never dump analysis to CLI.

Signals

GitHub stars
71
Forks
9
Last commit
Aug 2026
Advanced
Catalog kind
skill
Gateway key
deploy-recon
Source
github.com/tonone-ai/tonone