Upgrade Goose to v1.50.0 and enable GPT-6 Astra with goose-agent tool calling
Goose v1.50.0 adds GPT-6 Astra and goose-agent tool calling, requires HTTPS Snowflake, prefers latest MCP, and removes fast model routing plus the local-inference managed registry. Upgrade, pin explicit models, and update Snowflake, Databricks, MCP, and ACP clients.Why this loop
v1.50.0 changes how Goose selects models and talks to tools and data sources. Fast model routing and the managed model registry for local inference are gone, so workflows that auto-picked models will break until you pin GPT-6 Astra or another explicit model. goose-agent now supports tool calling, which is the path to use for agent tool use. Snowflake connections must be HTTPS. Kotlin Databricks callers can set the AI Gateway path. MCP clients should prefer the latest protocol version, and ACP is split into separate npm packages for the goose ACP client and acp binary. Session naming is bound to a trusted path, tracing honors content-capture opt-out, subagent platform guards are enforced, and permission revocations persist across managers—update shared Goose config so these constraints are the default rather than surprises.
Proposed actions
- Upgrade Goose to v1.50.0 from https://github.com/aaif-goose/goose/releases/tag/v1.50.0, then set the provider model to a GPT-6 Astra model supported in this release.
- Enable tool calling on goose-agent so agent sessions invoke tools through the new goose-agent tool-calling path instead of the previous no-tools agent.
- Stop using fast model routing and the managed model registry for local inference (both removed in v1.50.0); pin every Goose session and local-inference config to an explicit model id.
- Replace HTTP Snowflake URLs with HTTPS (Goose now requires HTTPS). For Kotlin Databricks AI Gateway callers, set the AI Gateway path in the Kotlin client. Prefer the latest MCP version, and install the separately published goose ACP client and acp npm packages.
- Name CLI sessions only on the trusted path Goose now binds to; keep tracing content-capture opt-out honored; expect subagent platform guards to block unsupported platforms and permission revocations to persist across managers, using the authoritative tool-approval details in the UI.
Agent prompt
Paste into your agent or query via MCP (get_agent_prompt) — free, no extra AI costGoose session task
DevAgentRadar → Goose
You are helping me adopt a real coding-assistant change. Work only from the facts below. Do not invent features.
Context
Assistant: Goose Proposal: Upgrade Goose to v1.50.0 and enable GPT-6 Astra with goose-agent tool calling Summary: Goose v1.50.0 adds GPT-6 Astra and goose-agent tool calling, requires HTTPS Snowflake, prefers latest MCP, and removes fast model routing plus the local-inference managed registry. Upgrade, pin explicit models, and update Snowflake, Databricks, MCP, and ACP clients. Primary source: https://github.com/aaif-goose/goose/releases/tag/v1.50.0
Why it matters
v1.50.0 changes how Goose selects models and talks to tools and data sources. Fast model routing and the managed model registry for local inference are gone, so workflows that auto-picked models will break until you pin GPT-6 Astra or another explicit model. goose-agent now supports tool calling, which is the path to use for agent tool use. Snowflake connections must be HTTPS. Kotlin Databricks callers can set the AI Gateway path. MCP clients should prefer the latest protocol version, and ACP is split into separate npm packages for the goose ACP client and acp binary. Session naming is bound to a trusted path, tracing honors content-capture opt-out, subagent platform guards are enforced, and permission revocations persist across managers—update shared Goose config so these constraints are the default rather than surprises.
Suggested actions
- Upgrade Goose to v1.50.0 from https://github.com/aaif-goose/goose/releases/tag/v1.50.0, then set the provider model to a GPT-6 Astra model supported in this release.
- Enable tool calling on goose-agent so agent sessions invoke tools through the new goose-agent tool-calling path instead of the previous no-tools agent.
- Stop using fast model routing and the managed model registry for local inference (both removed in v1.50.0); pin every Goose session and local-inference config to an explicit model id.
- Replace HTTP Snowflake URLs with HTTPS (Goose now requires HTTPS). For Kotlin Databricks AI Gateway callers, set the AI Gateway path in the Kotlin client. Prefer the latest MCP version, and install the separately published goose ACP client and acp npm packages.
- Name CLI sessions only on the trusted path Goose now binds to; keep tracing content-capture opt-out honored; expect subagent platform guards to block unsupported platforms and permission revocations to persist across managers, using the authoritative tool-approval details in the UI.
Config surfaces this release may change
- subagent definitions — check your repo before applying
- permission rules — check your repo before applying
- MCP servers — check your repo before applying
After you finish
Do not report this as applied to DevAgentRadar. You cannot write the visitor's loop.
Tell the human: open https://devagentradar.com/proposals/goose-v1-50-0-upgrade-goose-to-v1-50-0-and-enable-gpt-6-astra-with-goose and mark Applied, Skipped, or Failed. Proposal id: 1a99b774-002c-4640-bb31-eeac924c0aa9
Your job
- Restate the change in one sentence.
- Propose a minimal plan for my repo (or a throwaway pilot).
- Implement only what I approve; prefer small diffs and tests.
- Call out risks (permissions, breaking APIs, cost).
Start by confirming you understood the proposal.
Your loop
This browser · no sign-in · not shared as “you”After you run the prompt
Only you can mark this. Agents cannot write your loop.
Your decision stays on this device. A public tally appears after a few votes.