opencode

A set of project-wide rules for OpenCode, a coding-agent platform. It controls model defaults, provider configuration, and how OpenCode settings are maintained.

In plain words
What is it for?
Use it when managing OpenCode models, agents, providers, configuration synchronization, and cleanup of automatically saved provider entries.
Why use it?
It prevents incompatible or unwanted model settings from accumulating in project and live configuration files.

Cursor rule for Cursor

Install

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.

agentmods
npx agentmods add rules/wyattowalsh/agents/opencode
Clone the repo
git clone --depth 1 https://github.com/wyattowalsh/agents

Made for: Cursor.

Per session 0 Nothing until a file matches its globs; then the whole rule loads.
When invoked 3,039 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00000 $0.03039
Opus 5 $0.00000 $0.01520
Sonnet 5 $0.00000 $0.00608
Haiku 4.5 $0.00000 $0.00304

Measured yesterday against content hash 43f7f7f91c6e, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

opencode 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 yesterday.

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.

.cursor/rules/opencode.mdc · 164 lines

How it starts

The opening of the file, as written. The whole thing — 164 lines — stays where its author put it; the contents beside it link to each section on GitHub.

OpenCode Global Instructions

OpenCode-Specific Overrides

Model Preference

OpenCode model defaults are repo-managed. Repo and live OpenCode config set root model: "openai/gpt-5.5", root small_model: "openai/gpt-5.4-mini", and built-in agents build, plan, explore, and general to openai/gpt-5.5 with variant xhigh. Use the model plus variant fields for high-level thinking; do not invent composite model IDs such as openai/gpt-5.5-high unless the provider metadata defines that exact model. Do not generate OpenAI, Vercel, Kimi, or other remote provider blocks unless a concrete runtime need is verified. OpenCode's built-in provider registry should own normal model definitions; the only repo-managed exception is a minimal xai provider block with empty options so the desktop model picker can select Grok routes without inheriting incompatible OpenAI options. The sync tooling strips vercel/opencode-go/kimi-for-coding entries on every write of the project opencode.json and the live ~/.config/opencode/opencode.json, and sanitizes any xai.options back to {}. This prevents auto-persisted blocks for picker-chosen models from accumulating keys that the target provider rejects ("invalid xai provider options" etc.). Preserve explicit local provider blocks such as LM Studio only when rendered from local provider policy. Do not define duplicate gpt-5.5-fast entries unless concrete runtime evidence requires them. Do not silently fall back to another OpenAI model when CLI/TUI runs warn or retry; compare --pure, plugin-loaded, TUI, and desktop paths first and keep exact request IDs from any server_error lines.

Using Grok 4.3 (or other xai/vercel routes): Use the TUI model picker (<leader>m / model list + variant list). After exercising a new remote route, re-run the normal opencode sync (wagents or equivalent) so the tooling can scrub transient provider blocks and reduce provider.xai to empty options. The OpenAI block remains the only repo-managed provider entry with model variants and plugin options.

Read the full file on GitHub · 164 lines

Changes

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.

  1. yesterday First seen · 164 lines · 3,039 tokens per session scan A 43f7f7f91c6e

Subscribe to this mod's changes

opencode is a cursor rule published in the GitHub repository wyattowalsh/agents (5 stars, last pushed 11d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 3,039 tokens. 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 2026-08-31.