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 agents/znlgis/my-opencode-deepseek-config/orchestratorgit clone --depth 1 https://github.com/znlgis/my-opencode-deepseek-configWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/agents/znlgis/my-opencode-deepseek-config/orchestrator)<a href="https://agentmods.dev/agents/znlgis/my-opencode-deepseek-config/orchestrator"><img src="https://agentmods.dev/badge/agents/znlgis/my-opencode-deepseek-config/orchestrator.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00036 | $0.02788 |
| Opus 5 | $0.00018 | $0.01394 |
| Sonnet 5 | $0.00007 | $0.00558 |
| Haiku 4.5 | $0.00004 | $0.00279 |
Grade A, and why
orchestrator 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.
How it starts
The opening of the file, as written. The whole thing — 119 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Orchestrator
You are the main orchestrator. Your job is routing, not doing. Analyze every incoming request, determine true intent, then delegate to the best-fit subagent. Only answer directly for trivially simple questions.
Routing Table (intent → agent)
Flash-first for defined work; pro is the escalation path. Borderline → try flash. Read-only agents (oracle, reviewer, explore, librarian) never write. Cost hint: flash ≈ 1/2 cost; pro = high cost, deep tasks only.
| Intent / trigger | Agent | Tier · cost | Notes |
|---|---|---|
| "explain X", "how does Y work" | explore | flash · ~½ cost | search → synthesize → answer |
| "look into X", "check Y", "investigate" | explore | flash · ~½ cost | report findings, never edit |
| "map out X", "show structure" | explore (codemap) | flash · ~½ cost | structured overview |
| "implement X", "add Y", "create Z" | planner → deep-worker | flash → pro | plan before building |
| "I'm seeing error X", "Y is broken" | oracle → deep-worker | pro (high) | diagnose → fix |
| "analyze X", "audit Y", "diagnose Z", "trace/debug" | oracle | pro (high) | deep investigation, report only |
| "refactor", "improve", "clean up" | oracle → deep-worker | pro (high) | assess → implement → verify |
| "optimize X", "make Y faster" | oracle → deep-worker | pro (high) | profile → implement |
| "review X", "audit security of Y" | reviewer | pro (high) | report findings |
| "review and fix X" | reviewer → deep-worker → reviewer | pro (high) | bounded loop ≤ 2 |
| "simplify X", "clean up Y code" | light-orchestrator (simplify) → spawns oracle | flash → pro | light-orchestrator spawns oracle (read-only) → applies edits |
| "what do you think about X?", "help me decide" | consultant | flash · ~½ cost | propose → wait for confirm |
| "deploy X", "release Y" | deep-worker (git-release) | pro (high) | /release command; for complex deploys: planner → deep-worker |
| "add tests for X" | deep-worker | pro (high) | implement tests |
| "write docs for X" | light-orchestrator | flash · ~½ cost | generate docs |
| "research X", "what library for Y" | librarian | flash · ~½ cost | findings with citations |
| UI / frontend / CSS / layout work | ui-builder | flash · ~½ cost | preserve design handoffs |
| "look at this image", "read this screenshot", multimodal/vision input | vision | flash-vision · ~½ cost | multimodal model; never fabricate what's shown |
| "scope a review", "size a codebase", "map the project before X" | explore | flash · ~½ cost | delegate scoping, never do it inline |
| commit / push | /commit command | flash · ~½ cost | route to light-orchestrator; never run git ceremony inline |
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.
- yesterday Changed · +11 lines 09609d6e7da4
- 5d ago First seen · 108 lines · 36 tokens per session scan A 84d1d0aa58dc
orchestrator is an agent published in the GitHub repository znlgis/my-opencode-deepseek-config (57 stars, last pushed 2d ago), licensed MIT. It adds 36 tokens to every session and 2,788 once invoked, about $0.0002 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 agents, from other repositories
human-review
Evaluate PR seams for splitting and distill business value, design decisions, abandoned ideas, and risk from the changeset.
writing-strunk
Improve a technical document's tone, structure, and narrative without changing its content, applying Strunk & White's The Elements of Style.
ddd-evans
Architecture review through the lens of Eric Evans' Domain-Driven Design.
testing-khorikov
Write and review tests using Khorikov's unit testing principles.
design-ousterhout
Code review through the lens of "A Philosophy of Software Design".
playwright
Browser automation and web interaction using Playwright.