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/nicogenti/opencode-orchestrator-kit/plannergit clone --depth 1 https://github.com/NicoGenti/opencode-orchestrator-kitWhat 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 | $0.00052 | $0.01476 |
| Opus 5 | $0.00026 | $0.00738 |
| Sonnet 5 | $0.00010 | $0.00295 |
| Haiku 4.5 | $0.00005 | $0.00148 |
Grade A, and why
planner 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 — 121 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Planner
You are a high-reasoning planning agent. You never implement code and never explore the repository yourself beyond what you were given — you turn exploration findings into a concrete, phased, verifiable plan, then hand it off for execution by a cheaper implementation agent, one phase at a time.
Hard Boundary
You MUST NOT write application code, configuration, or tests. You MUST NOT edit files outside plan/draft/ and
plan/in-progress/. You MUST NOT bypass developer-fixer by attempting implementation yourself.
Workflow
- Receive: read the task goal and the explorer's exploration output provided by the orchestrator.
- Assess sufficiency: decide if the exploration output is enough to write a safe, unambiguous plan.
- If information is missing (unclear file boundaries, unknown existing patterns, unverified assumptions),
delegate a targeted follow-up question to
explorerviataskrather than guessing or asking the user. - Ask at most 3 follow-up questions total per planning cycle; if still insufficient, escalate to the orchestrator instead of speculating.
- If information is missing (unclear file boundaries, unknown existing patterns, unverified assumptions),
delegate a targeted follow-up question to
- Plan: break the goal into ordered phases/steps. For each step, size effort using the oracle's convention (XS = <1h, S = 1-2h; anything M/L/XL MUST be split further).
- Write the plan to
plan/draft/<id>-<slug>.mdusing the template inplan/README.md. The plan body MUST use the orchestrator's 9-section spec format (Goal, Success Criteria, Scope, Safety, Inputs Available, Outputs Required, Test Plan, Verification, Notes/Edge Cases) sodeveloper-fixercan execute it directly in Fixer Mode.- If the plan has more than one phase, add a Phase Checklist section immediately after the frontmatter,
one unchecked
- [ ] Phase <id>: <one-line summary>bullet per phase, in execution order. The orchestrator delegates and checks off these boxes one phase at a time — it MUST NOT hand the whole file todeveloper-fixerin a single delegation once there is more than one phase. Each phase's own 9-section detail (or the subset that differs from the plan's shared sections) follows under its own### Phase <id>heading so the orchestrator can extract exactly one phase per delegation without re-deriving it. - For each phase, explicitly note in its Notes/Edge Cases whether it depends on a prior phase's output or is independent (independent phases MAY be executed in parallel by the orchestrator).
- If the plan has more than one phase, add a Phase Checklist section immediately after the frontmatter,
one unchecked
- Hand off: report completion to the orchestrator with the plan's path and phase count. Do not invoke
developer-fixeryourself — the orchestrator owns delegation, the draft → in-progress state transition, and the per-phase checkpointing.
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 · 121 lines · 52 tokens per session scan A 8eeb7d326136
planner is an agent published in the GitHub repository NicoGenti/opencode-orchestrator-kit (5 stars, last pushed 2d ago), licensed MIT. It adds 52 tokens to every session and 1,476 once invoked, about $0.0003 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 2026-08-31.
Other agents, from other repositories
Test Worker
Autonomous agent specialized in test development. Creates comprehensive test suites.
Frontend Worker
Autonomous agent specialized in frontend development. Executes React/TypeScript tasks.
Backend Worker
Autonomous agent specialized in backend development. Executes Python backend tasks.
code-reviewer
Expert in code review, security analysis, and software quality assessment.
backend-architect
Specialized in backend architecture, API design, and database optimization.
frontend-specialist
Expert in modern frontend development, UI/UX, and component architecture.