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 skills add jamesdsizemore/rush-cli --skill skill-v2git clone --depth 1 https://github.com/jamesdsizemore/rush-cliWrote 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/skills/jamesdsizemore/rush-cli/skill-v2)<a href="https://agentmods.dev/skills/jamesdsizemore/rush-cli/skill-v2"><img src="https://agentmods.dev/badge/skills/jamesdsizemore/rush-cli/skill-v2/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/jamesdsizemore/rush-cli/skill-v2"><img src="https://agentmods.dev/badge/skills/jamesdsizemore/rush-cli/skill-v2.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00068 | $0.06390 |
| Opus 5.5 | $0.00027 | $0.02556 |
| Sonnet 5.5 | $0.00014 | $0.01278 |
| Haiku 4.5 | $0.00007 | $0.00639 |
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 10d 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 — 334 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Orchestrator
You are the orchestrator. You own the final product and the handoff to the user. Subagents do the work; you dispatch them, check what they return, and never hand over anything you have not verified. The mechanical steps (path check, task tests, staging, commit, status) are scripts; you spend your turns on judgment: dispatching, reading a 12-line result, deciding.
Absolute rules
These bind you and every subagent. No exceptions, no negotiation, no "good enough".
- No descoping. Every item in the plan and every item in the source spec gets done. Never
drop, defer, narrow, or merge away a task, requirement, finding, or test. Only the user can
remove scope, by an explicit statement in their own message.
[enforced:
orch-design-gatebriefs carry no open decisions;dispatch-prompt.shrefuses an implementer map withoutdone:lines;~/.claude/hooks/ask-guard.js] - No degrading. Never ship a weaker version of a task: no stubbed logic, faked output,
skipped or
xfailtests, loosened assertions,# type: ignore/noqato silence a real problem, or a "simplified for now" variant. [enforced: the Required clauses indispatch-prompt.sh; Check 4 oforch-reviewer;gate.sh(zero skips)] - No incomplete handoff. Handoff happens only when every task passed verification, every
review finding is fixed and re-verified, and docs are current. A blocker that needs the
user's decision is a grant parked in
.orchestrator/questions.mdwhile the other chains run; the run stays open. Context pressure is never a reason to stop: the run continues fromrun-state.sh report, the progress log, git log and the plan. A "status handoff" is an incomplete handoff. [enforced:run-state.sh stoprefuses with uncommitted changes;run-state.sh task mergedneedsgate-ok= HEAD] - Never Fable or Mythos. Neither you nor any subagent or external CLI may run on a Fable or
Mythos model.
- Step 0: read your own model from the system prompt ("You are powered by the model named
..."). If it is Fable or Mythos, stop and tell the user to run
/model opus. Do nothing else. - Dispatch only the
orch-*agents below; each pins its model by full id in frontmatter. Never usesubagent_type: "fork", never omitmodelfor a non-orch-*agent, never passmodel: "fable". Forks andmodel: inheritcopy the session model. - Never use the Workflow tool during a run (its agents can inherit the session model).
- Never pass a Fable/Mythos model to
codex,agy, orclaudeCLIs. [enforced:~/.claude/hooks/orchestrator-model-guard.js;probe-agents.shchecks each agent's model id at start] If the guard denies a call, fix the call; never route around the denial with another tool.
- Step 0: read your own model from the system prompt ("You are powered by the model named
..."). If it is Fable or Mythos, stop and tell the user to run
What ships with it
42 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- references/dispatch.md 8.7 KB
- scripts/decisions.sh 1.2 KB runs code
- scripts/disk-clean.sh 1.3 KB runs code
- scripts/dispatch-prompt.sh 15 KB runs code
- scripts/gate.sh 3.3 KB runs code
- scripts/load-plan.py 4.3 KB runs code
- scripts/monitor.py 21 KB runs code
- scripts/patch-agents.py 6.3 KB runs code
- scripts/probe-agents.sh 3.3 KB runs code
- scripts/run-state.sh 14 KB runs code
- scripts/save-map.sh 2.2 KB runs code
- scripts/task-gate.sh 2.6 KB runs code
- scripts/test-acceptance.py 6.0 KB runs code
- scripts/usage.py 6.9 KB runs code
- scripts/worktree-add.sh 935 B runs code
- tests/CHECKPOINT-rules.md 757 B
- tests/CHECKPOINT-scripts.md 1.3 KB
- tests/fixtures/agents/orch-adversarial-reviewer.md 2.9 KB
- tests/fixtures/agents/orch-docs.md 2.8 KB
- tests/fixtures/agents/orch-implementer.md 3.1 KB
- tests/fixtures/agents/orch-planner.md 3.6 KB
- tests/fixtures/agents/orch-reviewer.md 2.4 KB
- tests/fixtures/agents/orch-scout.md 2.2 KB
- tests/fixtures/agents/orch-verifier.md 2.4 KB
- tests/run.sh 414 B runs code
- tests/test_decisions.sh 1.3 KB runs code
- tests/test_disk_clean.sh 2.4 KB runs code
- tests/test_gate_lock.sh 2.3 KB runs code
- tests/test_gate.sh 2.9 KB runs code
- tests/test_load_plan.sh 2.5 KB runs code
- tests/test_model_guard_rule7.sh 2.6 KB runs code
- tests/test_monitor.py 21 KB runs code
- tests/test_patch_agents.sh 2.1 KB runs code
- tests/test_probe_agents.sh 2.8 KB runs code
- tests/test_run_state_graph.sh 7.1 KB runs code
- tests/test_run_state.sh 3.6 KB runs code
- tests/test_save_map.sh 2.2 KB runs code
- tests/test_skill_text.sh 2.7 KB runs code
- tests/test_task_gate.sh 2.4 KB runs code
- tests/test_test_acceptance.py 6.1 KB runs code
- tests/test_worktree_add.sh 2.3 KB runs code
- tests/test-dispatch-prompt.sh 14 KB runs code
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.
- 10d ago First seen · 334 lines · 68 tokens per session scan A 9a0cf3e361a2
orchestrator is a skill published in the GitHub repository jamesdsizemore/rush-cli (0 stars, last pushed today), licensed MIT. It adds 68 tokens to every session and 6,390 once invoked, about $0.0003 per session on Opus 5.5. 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 2026-10-05.
Other skills, from other repositories
local-ai-agents
Build local-first AI agents that run entirely on a developer workstation with Microsoft Foundry Local and Qwen function-calling models. Covers Small Language Models (SLMs), the OpenAI-compatible local endpoint, sandboxed local tools, local RAG with Chroma, local MCP servers, hybrid cloud/local routing, and the…
systematic-debugging
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.
next-cache-components-adoption
Turn on Cache Components in a Next.js app and resolve the blocking routes it surfaces. Use when the user wants to enable, adopt, or migrate to Cache Components, flip the cacheComponents flag, work through a flood of blocking-prerender / instant validation errors, run the cache-components-instant-false codemod, or…
chat-pet-sprite-creation
Use when creating or changing VS Code chat pet sprite art, sprite sheets, state animations, eye treatments, Stable/Insiders variants, or pet transitions under src/vs/workbench/contrib/chat/browser/widget/media/chatPet.
cpu-profile-analysis
Analyze V8/Chrome CPU profiles (.cpuprofile) and DevTools trace files (Trace-.json). Use when: profiling performance, investigating slow functions, comparing code paths, finding bottlenecks, analyzing timeToRequest, understanding call trees from sampling profiler data, analyzing layout/paint/rendering, investigating…
insight-error-page
Write or audit an insight-kind error page for the Next.js dev overlay. Use when creating a new errors/ .mdx page, auditing an existing one, or checking that a page matches the framework fix cards. Covers page structure, title alignment, FixCard cards with Copy prompt button, code snippets, terminology verification…