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.
npx skills add obto-inc/platform --skill obto-platform-boundarygit clone --depth 1 https://github.com/obto-inc/platformWrote 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.
[](https://agentmods.dev/skills/obto-inc/platform/obto-platform-boundary)<a href="https://agentmods.dev/skills/obto-inc/platform/obto-platform-boundary"><img src="https://agentmods.dev/badge/skills/obto-inc/platform/obto-platform-boundary/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/obto-inc/platform/obto-platform-boundary"><img src="https://agentmods.dev/badge/skills/obto-inc/platform/obto-platform-boundary.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00128 | $0.01605 |
| Opus 5 | $0.00064 | $0.00803 |
| Sonnet 5 | $0.00026 | $0.00321 |
| Haiku 4.5 | $0.00013 | $0.00161 |
Grade A, and why
obto-platform-boundary 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 5d 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.
How it starts
The opening of the file, as written. The whole thing — 71 lines — stays where its author put it; the contents beside it link to each section on GitHub.
The OBTO Platform Boundary
The invariant
The MCP surface is the whole platform. An OBTO client creates, deploys, diagnoses, verifies, and recovers applications using MCP tools and nothing else. You never need — and must never use or recommend — Kubernetes or kubectl, pod restarts or recycles, helm, container shells, SSH, a DNS console, a cloud provider console, direct database access, or any host filesystem path. If a tenant needed those, the platform would be broken, not the tenant.
This holds even when you have such access. A Claude Code session with a shell and a kubeconfig can reach an OBTO cluster; using it to make an app work is still wrong. It produces a result nobody can reproduce, silently mutates state the platform's own records do not know about, and hides a real product defect behind an operator's privileges. The tenant experience is the contract, and the tenant only has MCP.
Never tell a user to restart, recycle, redeploy, or reboot anything. Never emit kubectl, docker, helm, ssh, or DNS-console instructions in a response about making an OBTO app work.
What to do instead — by symptom
"I saved the code but the running app still behaves the old way." This is activation, and activation is the platform's job. Write receipts tell you where you stand:
read_per_request— live immediately, nothing to do.syncs_live— propagating across serving runtimes now; a probe may briefly hit an older copy.pending_platform_activation— durably stored, activating asynchronously.requires_site_rebuild— a site-served host serves the built bundle, so the record change is not live untilobto_build_apprebuilds it. This one has an MCP action and it is a rebuild, not a restart.
For pending_platform_activation: try obto_reload_scripts (it activates stored boot-loaded code where the serving runtime supports it, and refuses honestly where it does not). Otherwise verify rather than intervene — call something only the new code can answer (a new method, a new response field) with obto_invoke_route. Propagation is eventual and sometimes completes in minutes with no intervention at all. If it genuinely never activates, that is a platform incident: report the tool's error code to the operator. It is not a task you complete with infrastructure access.
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.
- 5d ago First seen · 71 lines · 128 tokens per session scan A 159bf806b275
obto-platform-boundary is a skill published in the GitHub repository obto-inc/platform (1 stars, last pushed 4d ago), licensed Apache-2.0. It adds 128 tokens to every session and 1,605 once invoked, about $0.0006 per session on Opus 5. 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 2026-09-04.
Other skills, from other repositories
agentcore-investigation
Investigate Bedrock AgentCore runtime sessions via CloudWatch Logs Insights — resolve session/trace IDs, query OTEL spans, filter noise, build timelines. Use when debugging AgentCore agent sessions, tracing tool calls, or analyzing latency.
n8n-self-hosting
Deploy a production self-hosted n8n end-to-end to a fresh Linux VM over SSH, using Docker Compose behind a Caddy reverse proxy with automatic HTTPS. Use whenever the user wants to self-host, install, set up, provision, or deploy n8n on their own server/VPS/box (Hetzner, DigitalOcean, AWS EC2, bare metal, etc.) — in…
cloud-run
Manage Cloud Run services and jobs.
zdx-investigate-multi-app-outage
Diagnose a multi-application outage scoped to one location by correlating ZDX alerts, affected devices, and shared cloud-path hops. Identifies the devices affected at a specific office, compares the per-application network path across multiple SaaS apps to surface the common network bottleneck, and produces an…
frontmcp-deployment
Use when deploying, building for production, packaging, or shipping a FrontMCP server. Covers build targets (node, cli SEA binary, browser, embeddable SDK, mcpb archive for Claude Desktop, serverless) and deploying to Vercel (with Vercel KV), AWS Lambda (API Gateway, SAM, CDK), Cloudflare Workers (KV, D1, Durable…
orchardcore-ai-documents-azure-storage
Skill for storing CrestApps Orchard Core AI Documents in Azure Blob Storage. Covers AIDocumentBlobStorageOptions, BlobFileStore replacement, tenant container lifecycle, cleanup choices, and Azure configuration. Use this skill when requests mention Orchard Core AI Documents Azure Blob Storage…