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.
npx agentmods add instructions/chryzxc/vscode-opencode/agents-mdgit clone --depth 1 https://github.com/chryzxc/vscode-opencodeWhat 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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00599 | $0.00599 |
| Opus 5 | $0.00300 | $0.00300 |
| Sonnet 5 | $0.00120 | $0.00120 |
| Haiku 4.5 | $0.00060 | $0.00060 |
Grade A, and why
vscode-opencode AGENTS.md 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 2d ago.
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.
How it starts
The opening of the file, as written. The whole thing — 65 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Agent Instructions
These rules apply to AI agents working in this repository.
Before editing
- Read the nearest
AGENTS.mdfiles for the files being changed. - Inspect
git status --shortand preserve unrelated worktree changes. - For streaming chat, activity timelines, hydration, deduplication, ordering, structured output, or collapsed/expanded response rendering, use the fixture or SDK debug data supplied with the issue. If none exists, add a minimal behavioral fixture before changing production code.
- Trace the full path: SDK event → normalization → store/merge → ownership and ordering → rendered component.
Required implementation behavior
- Add or update a regression test before changing streaming UI behavior.
- Rendered activity rows are append-only within an assistant turn. Later partial events may enrich rows but must not clear or replace already rendered rows without an explicit removal event.
- SDK transport IDs are not semantic visible-action identity. Do not dedupe distinct actions by title, timestamp, or mutable status.
- Preserve first-arrival order across assistant phases and hydration.
- A live response card and its transcript counterpart must have one response body owner; never render both copies for one assistant turn.
- Structured output must be selected and rendered in both live and hydrated paths. Collapsed and expanded views must choose the same final response.
- Do not add a fallback state owner or a new dedupe layer without a regression test covering the existing owner and merge path.
Validation
Run the focused tests first, then:
npm run guard:streaming
npm run typecheck
npm run webview:build
npm run build
Run npm run guard:prepush before handing off broad or cross-layer changes.
The streaming contract guard is authoritative. Do not weaken, remove, skip, or
rewrite a failing regression merely to make the guard pass.
Repository boundaries
src/is the extension host;webview/shared/is a separate Vite package.- Structured-output source files live in
src/shared/; update generated webview contracts throughnpm run structured-output:sync. - Keep transport-only prompt payloads free of wrapper-authored policy text.
- Do not remove existing user-facing controls or information density without explicit user direction.
- Never use
git restore,git checkout --,git clean, or reset commands to discard changes unless the user explicitly authorizes it.
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.
- 2d ago First seen · 65 lines · 599 tokens per session scan A a8bef3ba7bd8
vscode-opencode AGENTS.md is an instructions file published in the GitHub repository chryzxc/vscode-opencode (11 stars, last pushed 16d ago), licensed MIT. It adds 599 tokens to every session, about $0.0030 per session on Opus 5. 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 2026-08-30.
Other instructions, from other repositories
golembot CLAUDE.md
Instructions for 0xranx/golembot, covering golembot — project conventions for claude code, architecture hard constraints, things you must never do, interface change rules and file responsibility boundaries.
golembot AGENTS.md
Instructions for 0xranx/golembot, covering assistant context, installed skills, directory structure and conventions.
agents copilot-instructions.md
Instructions for benchflow-ai/agents, covering style guide, test style, bad and good.
awesome-copilot-id AGENTS.md
Instructions for GulajavaMinistudio/awesome-copilot-id, covering communication, explanation and documentation, markdown formatting, user communication style and workflow & methodology.
superset CLAUDE.md
Claude Code instructions for superset-sh/superset, a project described as: Superset is an agentic IDE to orchestrate 100+ coding agents in parallel. Run any agent with your own subscription.
give-me-job AGENTS.md
Instructions for kyoungbinkim/give-me-job, covering give-me-job repository instructions, default workflow, skill routing, working on this repository and development loop.