Initiative 7 of the strategy: turn every player that could otherwise eclipse Tenki into a channel. Co-marketing with the agent vendors (Anthropic, OpenAI, Cursor) via the Sandbox reference architectures isn't just integration — it's community capture (land inside their developer crowds) and harness-agnostic positioning (be the portable runtime the model-provider managed agents don't lock you into). Plus a GitHub Marketplace listing where developers already are; featured placement in AI-agent registries; the Luxor compute/enterprise tie-in; and broader AI-coding alliances. Each is framed below as a proposed target, with the threat it neutralizes and the asset it creates.
The agent vendors are simultaneously Tenki's biggest opportunity (their users need somewhere isolated to run code) and a latent threat (they could ship their own sandbox — the Layer-D disruption). Co-published reference architectures make Tenki the recommended runtime and a distribution surface the vendor sends users to. Sandbox already supports Claude Code and Codex — these are proposed co-marketing motions on top of working integrations.
This is community capture, not just integration. Each reference architecture lands Tenki inside an agent vendor's developer community — the Claude Code, Codex, and Cursor crowds are the densest, highest-intent top-of-funnel for the whole loop. The co-marketing motion is how Tenki captures those adjacent communities via the vendors themselves rather than competing for them cold.
And it's deliberately harness-agnostic. The same vendors are moving down the stack (Anthropic's Claude Managed Agents — Layer D — bundle model + sandbox + harness behind one API). Partnering up the stack, while staying MCP-native and running every harness, positions Tenki as the portable, controllable runtime that the model-provider managed agents don't lock you into — and the natural home for the long-running, self-improving agents Layer D underserves.
The flagship. Co-published "Claude Code + Tenki Sandbox" reference architecture: give your agent root without giving it yours. Joint blog post, Anthropic dev-day presence, Sandbox as a recommended runtime for Claude Code — and the harness-agnostic alternative to Claude Managed Agents for teams that want a portable, persistent runtime for long-running agents.
Threat neutralized: T4 (E2B/Modal/Daytona own sandbox mindshare) + the Layer-D managed-agent squeeze (partner up the stack instead of being collapsed out of it). Asset created: "Tenki is the reference safe-runtime for Claude Code" — and a foothold in the largest agent community.
Second first-class reference. "Codex + Tenki Sandbox" repo + 5-min video + cost/throughput calculator, co-marketed with OpenAI. Featured in Codex agent docs/examples where possible — capturing the Codex builder community on the same harness-agnostic footing.
Threat neutralized: T4 + Layer-D model-provider lock-in. Asset created: a maintained, vendor-blessed Codex runtime story, and a second agent community Tenki sits inside.
Highest-density AI-coding persona. Featured marketplace/integration placement, joint webinar, and a quickstart for running Cursor's agent inside Sandbox. Cursor users are the richest top-of-funnel for the whole loop — and the largest single adjacent community to capture.
Threat neutralized: T4 + Cursor Bugbot review overlap. Asset created: default-runtime relationship with the densest agent audience, captured via the harness rather than fought for cold.
The GitHub Marketplace listing is the single highest-leverage platform play — it puts Runners and Code Reviewer in front of exactly the developers whose CI bill Tenki cuts and whose PRs it reviews, on the platform they already live in. All statuses below are proposed targets, not shipped listings.
| Partner / surface | Integration | Threat → distribution | Proposed status |
|---|---|---|---|
| GitHub Marketplace | First-class listings for Runners + Code Reviewer | T2 — incumbent (GitHub) owns the surface → list inside it | Submit Phase 1 |
| Anthropic — recommended runtimes | Claude Code + Sandbox reference, featured in Anthropic channels | T4 — sandbox mindshare | Via Tier 1 alliance |
| OpenAI — Codex examples | Codex + Sandbox reference in agent docs/examples | T4 | Via Tier 1 alliance |
| Cursor Marketplace | Featured Sandbox/agent integration | T4 + review overlap | Submit Phase 2 |
| VS Code / Cline ecosystem | Quickstart + listing for the OSS agent crowd | O2 — agent explosion | Watch + apply |
| Vercel | Deploy/CI templates; Sandbox sample template | O6 — co-marketing | Phase 2–3 |
As the agent ecosystem fragments, developers increasingly find runtimes and tools through registries, awesome-lists, and directories. Featured Sandbox placement in each is cheap, compounding discovery — and turns the "agent explosion" (O2) into a long-tail acquisition channel rather than a threat.
Featured Sandbox placement in the major "tools for AI agents / coding agents" directories. Each listing is a discovery surface for builders choosing where their agent runs code.
Inclusion in the curated "awesome AI agents / agent infra" lists that developers treat as de-facto endorsements. Earned via the reference architectures and OSS posture.
Where Claude Code, Codex, Cursor, and Cline surface recommended runtimes/integrations, get Sandbox listed — closing the loop from Tier 1 alliances.
Where E2B / Modal / Daytona are compared, ensure Tenki Sandbox appears with the convergence + owned-compute story. Let evaluators self-select.
Tenki is the consumer-facing edge of Luxor's "compute as a commodity" thesis. The same developers who adopt Runners and Sandbox become the warm path into the /compute GPU marketplace and /hardware audience — and Luxor's profitable, SOC 1/2 Type II posture is enterprise-trust inheritance most dev-tools startups can't buy. This is a threat converted to distribution in reverse: instead of the crypto/mining association deterring developers, the owned-compute economics become the moat.
Position Tenki dev adoption as the top of the Luxor compute funnel: activated workspaces → enterprise compute conversations. Co-marketed enterprise/compute content for the /compute + /hardware audience.
Threat neutralized: T6 (margin/commoditization). Asset created: a structural cost story no VC-funded rival can match.
Warm enterprise references from Luxor's existing relationships (e.g., its crypto/infra network) — early design-partner and case-study candidates while the developer brand is built.
Asset created: named references from day one, feeding the cohort program.
Lead enterprise conversations with the compliance posture Tenki inherits — rare for a dev-tools startup, and a wedge against pure-plays in security-conscious accounts.
Threat neutralized: the "unknown vendor" objection (W8).
Beyond the agent vendors, the convergence loop (write → run CI → review) touches deploy, templates, and the wider AI-coding tooling world. Lightweight co-marketing alliances here extend reach without exclusivity — Tenki runs anywhere.
Deploy/CI templates and a Sandbox sample template. The AI-coding loop ends in a deploy; Vercel's agent-infra audience overlaps with Tenki's. Joint template + co-marketed walkthrough.
Why: closes the loop from "agent writes & runs code" to "ships it."
Co-marketing with adjacent AI-coding tools (review-adjacent, agent-orchestration, dev-environment) where there's a non-competing fit. Shared content, cross-references, joint webinars.
Why: compounding reach into the agent-builder persona.
Sponsor/co-present at Cursor, Anthropic, and AI-coding dev days (see Events). Pair with the cohort and hackathon programs to convert presence into activations.
Why: the agent-builder persona concentrates at these events.
Sequenced to the 12-month phasing. Lead with the partnerships that ride integrations Tenki already has (Claude Code, Codex) and the surface with the most concentrated buyers (GitHub Marketplace); save the broader alliances for once the references and benchmarks give Tenki something to co-market with.
The principle. A partnership is only worth pursuing if it converts a specific ecosystem threat into a measurable distribution channel. We start where the integration already works and the audience is densest — then expand outward as the developer brand earns the right to ask for more.
| Priority | Partnership | Why first | Phase |
|---|---|---|---|
| 1 | Anthropic (Claude Code) + OpenAI (Codex) co-marketing | Integrations already ship; the reference architectures are Initiative 2's flagship deliverable; biggest threat (T4) neutralized fastest | Phase 1–2 |
| 2 | GitHub Marketplace listing (Runners + Code Reviewer) | Highest concentration of in-market buyers; rides the GitHub-pricing moment; pure distribution, low partnership cost | Phase 1 |
| 3 | Cursor (Anysphere) | Densest agent audience; marketplace + joint webinar; one more vendor-blessed runtime story | Phase 2 |
| 4 | AI-agent registries & directories | Cheap, compounding discovery; turns the agent explosion into a long-tail channel; can run continuously | Phase 1–3 (ongoing) |
| 5 | Luxor compute / enterprise tie-in | Unlocks enterprise expansion + the cost-moat narrative once the developer funnel is producing activated workspaces | Phase 3 |
| 6 | Broader AI-coding alliances (Vercel, etc.) | Extends reach after the references + benchmarks give Tenki co-marketing assets; nice-to-have, not load-bearing | Phase 2–3 |