custody-dataplane

custody-dataplane is a skill for Claude Code, Codex from lukasrepublic/agentic-foundry. It costs 0 tokens per session (1,020 once invoked), scanned A, original, MIT.

A procedure for isolating a sensitive Kubernetes workload with a separate security group for each pod. A security group is a network firewall rule set; the procedure also covers the cluster and network conditions required for it to work.

In plain words
What is it for?
Use it when introducing per-pod security groups, configuring the required cluster permissions and worker nodes, troubleshooting branch network interfaces, and restricting data-tier access to approved pods.
Why use it?
It helps explain why protected pods remain pending or why all their outbound connections time out when required cloud permissions, node support, or network settings are missing.

Skill for Claude CodeCodex

Part of the foundry plugin — 80 skills, 8 agents, 8 hooks, 1 MCP server shipped together

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/lukasrepublic/agentic-foundry/custody-dataplane
Any agent
npx skills add lukasrepublic/agentic-foundry --skill custody-dataplane
Clone the repo
git clone --depth 1 https://github.com/lukasrepublic/agentic-foundry

Made for: Claude Code, Codex.

Or install foundry, the plugin that ships this one along with the rest of its 80 skills, 8 agents, 8 hooks, 1 MCP server.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for custody-dataplane

README.md
[![agentmods](https://agentmods.dev/badge/skills/lukasrepublic/agentic-foundry/custody-dataplane.svg)](https://agentmods.dev/skills/lukasrepublic/agentic-foundry/custody-dataplane)
Your own site
<a href="https://agentmods.dev/skills/lukasrepublic/agentic-foundry/custody-dataplane"><img src="https://agentmods.dev/badge/skills/lukasrepublic/agentic-foundry/custody-dataplane.svg" alt="Measured on agentmods" height="20"></a>
Per session 0 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,020 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00000 $0.01020
Opus 5 $0.00000 $0.00510
Sonnet 5 $0.00000 $0.00204
Haiku 4.5 $0.00000 $0.00102

Measured 4d ago against content hash 75adf569631f, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

custody-dataplane scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

packs/stack-profiles/aws-eks-karpenter/blueprints/custody-dataplane/SKILL.md · 65 lines

How it starts

The opening of the file, as written. The whole thing — 65 lines — stays where its author put it; the contents beside it link to each section on GitHub.

When to trigger

  • Isolating a custodial/sensitive workload with per-pod security groups (security-groups-for-pods / branch ENI) so the data tier admits ONLY the dedicated pod SG.
  • A SecurityGroupPolicy pod stuck Pending forever, or a branch-ENI pod whose every outbound connection times out.
  • Reviewing the first per-pod-SG workload on a cluster (the enablers are cluster-wide substrate — the workload author is the least likely to notice they're absent).

The five silent enablers (ALL must hold; every one fails silently)

  • 1 — The VPC-resource-controller policy on the CLUSTER role. The base cluster policy does NOT grant trunk/branch-ENI creation. Missing ⇒ TrunkENICreationFailed: UnauthorizedOperation, no node advertises pod-eni capacity, and SGP pods stay Pending forever.
  • 2 — Trunk-CAPABLE worker nodes. A node advertises branch-ENI capacity only if it registered AFTER the IAM policy was live AND ran the CNI with pod-ENI enabled. Pre-existing/bootstrap nodes do NOT retro-attach a trunk — cycle the node or restart its CNI agent so it re-registers. Karpenter-provisioned fresh nodes pick it up automatically once enabler 1 is live.
  • 3 — SNAT exclusion for the data/peer CIDRs. Default CNI SNAT sends pod→outside-VPC traffic out the node primary ENI, which carries the NODE SG, not the pod SG — a data-tier SG admitting only the pod SG silently drops it (connection timeout). Exclude the peered/data VPC CIDRs from SNAT so the pod SG survives the hop.
  • 4 — Node-SG :53 ingress FROM the pod SGs. Cluster DNS sits behind the node SG, whose :53 ingress typically admits only the node SG itself — the branch-ENI pod's DNS queries (sourced from the pod SG) are dropped and resolution hangs. The most-missed enabler: DNS "already works" for every non-branch-ENI pod, and the symptom surfaces as a data-tier timeout.
  • 5 — Verify the pod SG's LIVE egress rules were APPLIED — not just authored. An SG that exists and a CR that references it look correct even when the egress rules never landed (a partial apply, a provider that dropped the rule). Assert the LIVE egress rules match intent — distinct from confirming the SG and CR exist.

Read the full file on GitHub · 65 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 4d ago First seen · 65 lines · 0 tokens per session scan A 75adf569631f

Subscribe to this mod's changes

custody-dataplane is a skill published in the GitHub repository lukasrepublic/agentic-foundry (1 stars, last pushed 3d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 1,020 tokens. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.