# Weekstart Brief — Issue 1 **Product:** Weekstart by Stackyard · **Domain:** stackyard.fyi **Date:** Week of 2026-09-13 · Sunday release **Tagline:** Get your agent stack set straight before the week starts. --- ## TL;DR - **Cost + compliance first:** OpenAI Agents API has no platform fee, but hosted sandboxes bill by the minute — and Agents is **US residency only / no ZDR**, even on self-hosted sandboxes. - **Still broken if you lagged:** Assistants API is dead (Aug 26). Fable/Mythos 5.1 still 400 on `tool_choice: any|tool`. - **Pin your defaults:** Gemini Antigravity now defaults to `gemini-3.8-flash`; Cursor Projects burn Cloud Agent / plan pools (no separate Projects quota). - **Wire carefully:** Anthropic MCP tunnels moved to `/v1/tunnels`; Google IAM UAP can deny MCP tools at the gateway; LangChain Deep Agents 0.7 changes todos + filesystem outputs. --- ## What changed ### 1. OpenAI Agents API — sandbox billing + ZDR/US limits · #breaking No separate Agents platform fee. Hosted sandboxes bill at container rates (1/4/16/64 GB → $0.03/$0.12/$0.48/$1.92 per 20-min session unit), eligible sessions **per minute with a 5-min minimum**. Model + tool fees extra. Agents API is **US data residency only** and **does not support ZDR** — self-hosted does **not** make it ZDR-eligible. **Why it matters:** “No Agents fee” is not free compute, and compliance assumptions from Chat/Completions won’t carry over. **Action:** Pin container tier, reuse long-lived containers, gate Agents API behind US-residency / non-ZDR policy before prod. Wait for environment `connected` before file ops (1h idle expiry on hosted). **Links:** [Agents overview](https://developers.openai.com/api/docs/guides/agents-api/overview) · [Hosted environments](https://developers.openai.com/api/docs/guides/agents-api/environments/openai-hosted) · [Pricing](https://developers.openai.com/api/docs/pricing) ### 2. Assistants API is still dead — finish the cutover · #breaking Sunset **2026-08-26**. Assistants→Prompts, Threads→Conversations, Runs→Responses. No automated Threads→Conversations migration. **Why it matters:** Lagging SDK paths fail hard Monday morning. **Action:** Cut over to Responses + Conversations (or Agents API for durable work). Rebuild history from app-stored messages. Prefer prompt specs in source control — dashboard prompt objects are on a deprecation path too. **Links:** [Migration guide](https://developers.openai.com/api/docs/assistants/migration) · [Deprecations](https://developers.openai.com/api/docs/deprecations) ### 3. Gemini Antigravity default → `gemini-3.8-flash` · #breaking Managed-agent default moved to **Gemini 3.8 Flash**. Saved agents lock model at `agents.create` (interaction-time override blocked). **Why it matters:** Unpinned runs silently changed cost and behavior. **Action:** Set `agent_config.model` explicitly on every create; audit existing agents; set `max_total_tokens` to stop runaway loops. **Links:** [Antigravity agent docs](https://ai.google.dev/gemini-api/docs/antigravity-agent) · [Managed agents expansion](https://blog.google/innovation-and-ai/technology/developers-tools/expanding-managed-agents-gemini-api-3-6-flash-hooks/) ### 4. Cursor Projects — no separate quota · #breaking Projects burn **Cloud Agent / Cursor Models / Other Models** pools. Subscriptions and automations are always-on meters. **Why it matters:** Parallel subagents can exhaust the month and trip on-demand charges with no “Projects limit” line item. **Action:** Set spend limits before Monday rollouts; watch the Spending tab; prefer Cursor Models pool where possible. **Links:** [Projects](https://cursor.com/blog/projects) · [Models & pricing](https://cursor.com/docs/models-and-pricing) · [Usage limits](https://prod.cursor.com/help/models-and-usage/usage-limits) ### 5. LangChain Deep Agents v0.7 — lean harness, breaking defaults ~65% fewer base input tokens; todos opt-in; backend factories removed; `ls`/`glob`/`read_file` output shapes changed; `delete` added to default FS tools. **Why it matters:** Teams on ≤0.6 will see missing todos and broken parsers after upgrade. **Action:** Upgrade with a migration checklist — restore `TodoListMiddleware` if needed; pass concrete backends + explicit `StoreBackend(namespace=...)`; re-parse FS tool outputs; allowlist FS tools if you don’t want `delete`. **Links:** [Deep Agents v0.7](https://www.langchain.com/blog/deep-agents-v0-7) · [Changelog](https://docs.langchain.com/oss/python/releases/changelog) · [Release](https://github.com/langchain-ai/deepagents/releases/tag/deepagents%3D%3D0.7.0) ### 6. Platform refresh (after the spine) - **Anthropic MCP tunnels → `/v1/tunnels`** (research preview) — migrate off Admin `/v1/organizations/tunnels` — [overview](https://platform.claude.com/docs/en/agents-and-tools/mcp-tunnels/overview) - **Managed Agents `auto` permissions** (carry-forward) — evaluation events on each tool call — [permission policies](https://platform.claude.com/docs/en/managed-agents/permission-policies) - **Gemini hooks + token budgets** — `.agents/hooks.json` pre/post tool gates + `max_total_tokens` — [hooks post](https://blog.google/innovation-and-ai/technology/developers-tools/expanding-managed-agents-gemini-api-3-6-flash-hooks/) - **Google IAM UAP for Agent Gateway ↔ MCP** — CEL allow/deny on tool names; start `DRY_RUN` then `ENFORCE` — [UAP overview](https://docs.cloud.google.com/gemini-enterprise-agent-platform/govern/policies/iam-overview-uap) - **Fable/Mythos 5.1 `tool_choice`** (carry-forward) — `any`/`tool` → 400; use strict/structured — [release notes](https://docs.anthropic.com/en/release-notes/api) --- ## Data structures & schemas ### Agents API environment shape Choose `environment.type: openai_hosted` vs `self_hosted`; network `enabled|disabled|restricted`; packages / setup_commands / files. Header: `OpenAI-Beta: agents=v1`. ### Anthropic MCP tunnels Beta `mcp-tunnels-2026-06-22`; WIF scope `workspace:manage_tunnels`. IDs: `tnl_*`, `tcrt_*`. Max 2 active CAs. Admin API keys rejected. Pair Messages/Managed Agents with tunneled URL + `mcp-client-2025-11-20`. ### Gemini Antigravity agent_config `{ type: "antigravity", model, max_total_tokens }`. MCP tools: `type: "mcp_server"`, lowercase alphanumeric `name`, Streamable HTTP only. Hooks: matcher regex + deny JSON `{"decision":"deny","reason":"..."}`. ### Deep Agents 0.7 filesystem / store Removed `BackendFactory` / `BACKEND_TYPES` / `FileFormat` / `Unset`. `FileData.content` is a string. Empty `ls`/`glob` → `"No files found"`. No `cat -n` gutter on reads. Pass explicit `StoreBackend(namespace=...)`. ### Claude Fable/Mythos 5.1 tools Prefer **strict tool use** or **structured outputs** (`output_config.format`). `tool_choice` only `auto`/`none`. Thinking blocks are model-bound on replay. ### Google UAP ↔ MCP CEL on `destination.agent_registry.mcp_server.name`, `.tool.name`, `.tool.annotations.destructive_hint` / `read_only_hint`. Principals as SPIFFE agent identities. Deny overrides allow. No VPC-SC for this feature. --- ## Tool & API watch 1. **OpenAI Agents API** — hosted vs self-hosted sandboxes, per-minute billing, ZDR/US limits — [overview](https://developers.openai.com/api/docs/guides/agents-api/overview) 2. **Anthropic MCP tunnels** — `/v1/tunnels` + cert/token lifecycle — [reference](https://platform.claude.com/docs/en/agents-and-tools/mcp-tunnels/reference) 3. **Managed Agents `auto`** — `evaluated_permission` + `evaluation.reason_code` — [policies](https://platform.claude.com/docs/en/managed-agents/permission-policies) 4. **Gemini Antigravity + hooks** — model pin, hooks, Environments API — [docs](https://ai.google.dev/gemini-api/docs/antigravity-agent) 5. **Google IAM UAP** — dry-run then enforce on Agent Gateway egress — [manage policies](https://docs.cloud.google.com/gemini-enterprise-agent-platform/govern/policies/manage-iam-policies-uap) 6. **Deep Agents 0.7** — todos opt-in, FS tool allowlists — [blog](https://www.langchain.com/blog/deep-agents-v0-7) 7. **Cursor Projects spend** — pools + on-demand — [pricing](https://cursor.com/docs/models-and-pricing) *Research note: post–Batch-1 window was thin — six watch-driven new finds + four carry-forward must-patches.* --- ## Operator checklist — get set straight **Monday must-patch** - [ ] **Agents API cost/compliance:** Confirm container tier + US-residency / non-ZDR policy before any prod Agents session. - [ ] **Assistants greps:** Kill leftover Assistants SDK/API usage; Responses/Conversations or Agents only. - [ ] **Gemini model pin:** Every Antigravity/managed agent has explicit `agent_config.model` + `max_total_tokens`. - [ ] **Projects spend:** Caps on before parallel Project subagents; check Spending tab. - [ ] **Fable/Mythos tools:** No `tool_choice: any|tool`; strict/structured instead. - [ ] **MCP tunnels:** Migrate to `/v1/tunnels` if still on org Admin tunnels path. **Then** - [ ] Deep Agents ≤0.6 → 0.7 migration checklist (todos, backends, FS parsers). - [ ] Managed Agents: `auto` only where irreversible side effects are acceptable; keep `always_ask` on bash/destructive. - [ ] Gemini: ship `.agents/hooks.json` for write/exec gates; clean Environments before 7-day TTL. - [ ] Google UAP: `DRY_RUN` CEL policies on MCP tool names, then `ENFORCE`. - [ ] Schema gate + least privilege on your single most important agent before Wednesday experiments. --- ## How to use this issue 1. Skim **TL;DR** (30 seconds). 2. Hit every **#breaking** item before Wednesday experiments. 3. Run the checklist; check off what you finish. 4. Hand `issue-001-agent-pack.json` to your bot if you want the same facts machine-side — **do not auto-execute** `action` fields. 5. Keep primary links open for production cuts. ## Pricing (Stackyard) | Plan | Price | Includes | |------|-------|----------| | Operator | $10/mo | Human brief, Monday checklist, schema & tool watch | | Operator + Agents | $20/mo | Everything in Operator + JSON pack + agent-ingest notes | Get set straight: [stackyard.fyi](https://stackyard.fyi) --- ## Agent-ingest notes Companion pack: **`briefs/issue-001-agent-pack.json`** 1. Read `issue` + `summary` first; surface `summary` unedited. 2. Prefer `priority: must_patch` / `#breaking` for Monday digests. 3. Group `items[]` by `category`; treat `action` as suggestion — never auto-execute. 4. Follow `links[]` to primary docs only. 5. Source handoff: `/workspace/stackyard/research/source-pack-issue-001-2026-09-13.md` --- ## Footer **Weekstart Brief · Issue 1** · Stackyard · stackyard.fyi Verified deltas for operators who run agents. Research → Editor → Ops. **Pipeline:** Research → Editor → Blaze review → Ops publish → Growth after go.