Borrowing it
Nothing to install: this file belongs to qingshihuan/qing-agent-orchestrator. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/qingshihuan/qing-agent-orchestrator/main/.agents/skills/qing-agent-orchestrator-full/SKILL.mdgit clone --depth 1 https://github.com/qingshihuan/qing-agent-orchestratorWrote 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/qingshihuan/qing-agent-orchestrator/qing-agent-orchestrator-full)<a href="https://agentmods.dev/skills/qingshihuan/qing-agent-orchestrator/qing-agent-orchestrator-full"><img src="https://agentmods.dev/badge/skills/qingshihuan/qing-agent-orchestrator/qing-agent-orchestrator-full/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/qingshihuan/qing-agent-orchestrator/qing-agent-orchestrator-full"><img src="https://agentmods.dev/badge/skills/qingshihuan/qing-agent-orchestrator/qing-agent-orchestrator-full.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 | $0.00036 | $0.00780 |
| Opus 5 | $0.00018 | $0.00390 |
| Sonnet 5 | $0.00007 | $0.00156 |
| Haiku 4.5 | $0.00004 | $0.00078 |
Grade A, and why
qing-agent-orchestrator-full 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 4d 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 — 27 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Qing
Reclassify the remaining phase each user turn and before delegation, revision or review:
- Direct: safe single-scope work stays with the parent; zero children, no model allocation or Handoff approval.
- Lite: one Executor, targeted parent verification, at most one revision.
- Full: high-risk effects, cross-system work, real parallelism, release/deploy or explicit Full/Level 3; structured Handoff and independent Reviewer, two children and one revision by default. Only explicit bounded configuration overrides these limits.
De-escalate as work narrows. Retain pending independent-review obligations for unaccepted high-risk artifacts; restore Full at final acceptance. PARENT_ACTION_REQUIRED and PARENT_VERIFICATION_REQUIRED are not completion; never fabricate a Review.
Conditional references
For clear Direct/Lite, do not preload references or schemas. Ambiguous routing or Full loads orchestrator-spec.md. Destructive, external, secret, global/system or uncertain effects require safety-gates.md and exact approvals before acting. Reversible in-scope edits/builds/tests need no plan approval. Preserve unrelated work; re-gate changed effects or scope. Full also loads handoff-protocol.md, reviewer-rules.md, Handoff schema and Review schema. Never skip required review or verification to save tokens.
Context and work budget
Delegate only useful separate work, never duplicate exploration. Send objective, relevant file ranges, constraints, acceptance criteria and tests, not the full conversation/repository. Revisions send unresolved findings and changed facts while retaining the safety contract. Reuse verification only for the same unchanged inputs, commands and environment; retest affected changes and run required full checks before release. Astra is opt-in for demanding work, not Direct/Lite's default. Choose an adequate capability-valid pair; verify current host model/effort availability. Use high for an initial demanding Astra choice; xhigh/max need a concrete reason. Keep parent model unchanged. Before spawning show ownership, role/task and actual pair (or inherited/unconfirmed). Use only valid explicit same-backend fallbacks; disclose actual substitutions, not unused candidates. Wait for meaningful events/results, not unchanged polling. One genuine no-progress timeout permits one takeover/replacement decision, not interrupt/reactivate loops. Report outcome, executionOwner (ChatGPT or Codex), artifacts, verification and limits. Plans, mocks and heartbeats are not completion; no per-tool ledger.
What ships with it
46 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.
- agents/openai.yaml 313 B
- references/codex-exec.md 1.6 KB
- references/execution-modes.md 1.2 KB
- references/handoff-protocol.md 980 B
- references/orchestrator-spec.md 3.8 KB
- references/reviewer-rules.md 1.1 KB
- references/safety-gates.md 1.2 KB
- runtime/config/relay.user.json 858 B
- runtime/dist/src/classifier.js 2.0 KB runs code
- runtime/dist/src/cli.js 33 KB runs code
- runtime/dist/src/codex-model-catalog.js 5.0 KB runs code
- runtime/dist/src/codex-runtime.js 12 KB runs code
- runtime/dist/src/config.js 16 KB runs code
- runtime/dist/src/execution-mode-router.js 6.1 KB runs code
- runtime/dist/src/executors/codex-exec-executor.js 24 KB runs code
- runtime/dist/src/executors/dry-run-executor.js 848 B runs code
- runtime/dist/src/executors/executor.js 11 B runs code
- runtime/dist/src/executors/mock-executor.js 1.3 KB runs code
- runtime/dist/src/git-audit.js 12 KB runs code
- runtime/dist/src/model-health.js 16 KB runs code
- runtime/dist/src/model-router.js 14 KB runs code
- runtime/dist/src/orchestration-policy.js 13 KB runs code
- runtime/dist/src/output-efficiency.js 2.0 KB runs code
- runtime/dist/src/planner.js 13 KB runs code
- runtime/dist/src/process-runner.js 10 KB runs code
- runtime/dist/src/relay-criterion-collector.js 3.9 KB runs code
- runtime/dist/src/relay.js 27 KB runs code
- runtime/dist/src/reviewer.js 5.4 KB runs code
- runtime/dist/src/run-store.js 28 KB runs code
- runtime/dist/src/safety-gate.js 6.2 KB runs code
- runtime/dist/src/sandbox-preflight.js 4.4 KB runs code
- runtime/dist/src/task-analyzer.js 2.9 KB runs code
- runtime/dist/src/task-router.js 17 KB runs code
- runtime/dist/src/types.js 11 B runs code
- runtime/dist/src/validation.js 24 KB runs code
- runtime/dist/src/workspace.js 1.5 KB runs code
- runtime/package.json 1.8 KB
- runtime/schemas/criterion-evidence.schema.json 932 B
- runtime/schemas/executor-result.schema.json 2.3 KB
- runtime/schemas/handoff.schema.json 6.8 KB
- runtime/schemas/model-health.schema.json 438 B
- runtime/schemas/model-route.schema.json 16 KB
- runtime/schemas/planner-output.schema.json 4.5 KB
- runtime/schemas/review.schema.json 2.6 KB
- runtime/schemas/test-evidence.schema.json 1.4 KB
- scripts/qing.ps1 1.3 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.
- 4d ago Changed · -6 lines · -47 tokens per session d7e8281535c7
- 12d ago First seen · 33 lines · 83 tokens per session scan A b4293cb23629
qing-agent-orchestrator-full is a skill published in the GitHub repository qingshihuan/qing-agent-orchestrator (2 stars, last pushed 4d ago), licensed MIT. It adds 36 tokens to every session and 780 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 2026-08-31.
Other skills, from other repositories
orchestrate
Plan and supervise user-approved software initiatives through Codex native tasks and bounded native subagents. Use when the user explicitly asks Agent Orchestration Gateway to clarify work, split it into non-overlapping modules, dispatch module tasks, or run an AOG module task.
orcast
Use when orchestrating multiple agents in Orca: fan out parallel workers, pipeline work across stages, adversarially verify findings, run judge panels or loop-until-dry sweeps, or delegate a task and get the result back. You write a Plan definition in real JavaScript — a worker-workflow graph of worker, action, and…
simply-pm
Use when you are spawned as the PM of a simply harness project and your brief references a .simply/LEDGER.md work ledger or this skill. Grants fleet authority, routing doctrine, safe worker lifecycle, and the signal protocol back to the auditor.
improve
Use ONLY when the user explicitly asks to optimize a measurable quantity of an existing system (performance, size, score, latency) and a trustworthy metric exists or can be justified. Most work does not need this; the standard criteria loop covers it.
sonol-multi-agent
Sonol multi-agent orchestration skill. Use when a user wants one request decomposed into single-agent or multi-agent execution, with local-AI creative draft generation, hosted normalization/binding, adapter-backed sub-agent controls, runtime reporting, dashboard monitoring, plan/run persistence, and final result…
sonol-agent-runtime
Sonol runtime helper skill for sub-agents participating in Sonol multi-agent orchestration. Use when a sub-agent must report progress, artifacts, completion, or session status into the shared dashboard and SQLite-backed run history.