Upgrade OpenCode to v1.18.17 and wire Copilot PDF plus stable long sessions
Ship OpenCode v1.18.17 so session compaction keeps full recent turns, retries stop storming, Copilot PDFs attach when supported, and DeepSeek V4 Flash plus Muse models get correct defaults and prompts.Why this loop
v1.18.17 changes long-session behavior: compaction keeps complete recent turns and writes clearer summaries for smaller models, and automatic session retries are capped with jitter so failures stop retry-storming. Copilot models that advertise PDF vision can take PDF attachments. DeepSeek V4 Flash gets the correct sampling defaults on supported providers, every Muse-family model is routed to the Meta system prompt, and MERGE Gateway reasoning variants are registered so those options resolve. Default-model selection now uses the server’s current default instead of stale local config. Shipping this tag is how those session, attachment, and routing fixes land.
Proposed actions
- Upgrade the OpenCode install on this machine to exactly v1.18.17 from https://github.com/anomalyco/opencode/releases/tag/v1.18.17 and run the CLI version command until it prints 1.18.17; remove any older opencode binary still on PATH.
- In OpenCode, pick a GitHub Copilot model that advertises PDF vision support, attach a real PDF, and confirm the request includes the PDF; do not attach PDFs to Copilot models that do not advertise PDF input.
- Drive a session long enough to trigger compaction and confirm the compacted context still contains complete recent turns plus the clearer small-model summary; then induce a transient session error and confirm automatic retries stop after the new cap and are jittered instead of immediate retry storms.
- Select DeepSeek V4 Flash on each supported provider and confirm this release’s sampling defaults are applied; then select every Muse-family model and confirm each is routed to the Meta system prompt.
- Turn on MERGE Gateway reasoning-variant model options and confirm those variants resolve and run instead of failing model lookup; also confirm new sessions pick the server’s current default model rather than a stale local default.
Agent prompt
Paste into your agent or query via MCP (get_agent_prompt) — free, no extra AI costOpenCode session task
DevAgentRadar → OpenCode
You are helping me adopt a real coding-assistant change. Work only from the facts below. Do not invent features.
Context
Assistant: OpenCode Proposal: Upgrade OpenCode to v1.18.17 and wire Copilot PDF plus stable long sessions Summary: Ship OpenCode v1.18.17 so session compaction keeps full recent turns, retries stop storming, Copilot PDFs attach when supported, and DeepSeek V4 Flash plus Muse models get correct defaults and prompts. Primary source: https://github.com/anomalyco/opencode/releases/tag/v1.18.17
Why it matters
v1.18.17 changes long-session behavior: compaction keeps complete recent turns and writes clearer summaries for smaller models, and automatic session retries are capped with jitter so failures stop retry-storming. Copilot models that advertise PDF vision can take PDF attachments. DeepSeek V4 Flash gets the correct sampling defaults on supported providers, every Muse-family model is routed to the Meta system prompt, and MERGE Gateway reasoning variants are registered so those options resolve. Default-model selection now uses the server’s current default instead of stale local config. Shipping this tag is how those session, attachment, and routing fixes land.
Suggested actions
- Upgrade the OpenCode install on this machine to exactly v1.18.17 from https://github.com/anomalyco/opencode/releases/tag/v1.18.17 and run the CLI version command until it prints 1.18.17; remove any older opencode binary still on PATH.
- In OpenCode, pick a GitHub Copilot model that advertises PDF vision support, attach a real PDF, and confirm the request includes the PDF; do not attach PDFs to Copilot models that do not advertise PDF input.
- Drive a session long enough to trigger compaction and confirm the compacted context still contains complete recent turns plus the clearer small-model summary; then induce a transient session error and confirm automatic retries stop after the new cap and are jittered instead of immediate retry storms.
- Select DeepSeek V4 Flash on each supported provider and confirm this release’s sampling defaults are applied; then select every Muse-family model and confirm each is routed to the Meta system prompt.
- Turn on MERGE Gateway reasoning-variant model options and confirm those variants resolve and run instead of failing model lookup; also confirm new sessions pick the server’s current default model rather than a stale local default.
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/opencode-v1-18-17-upgrade-opencode-to-v1-18-17-and-wire-copilot-pdf-plus and mark Applied, Skipped, or Failed. Proposal id: 08868697-c062-4e9f-aa1b-de6ec73cc738
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.
modelRelease source ↗
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.