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 jellydn/my-ai-tools --skill orchestrating-fusiongit clone --depth 1 https://github.com/jellydn/my-ai-toolsWrote 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/jellydn/my-ai-tools/orchestrating-fusion)<a href="https://agentmods.dev/skills/jellydn/my-ai-tools/orchestrating-fusion"><img src="https://agentmods.dev/badge/skills/jellydn/my-ai-tools/orchestrating-fusion/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/jellydn/my-ai-tools/orchestrating-fusion"><img src="https://agentmods.dev/badge/skills/jellydn/my-ai-tools/orchestrating-fusion.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 1 finding, up to medium
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- medium Rogue Agent · line 22 Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.Fix: Remove any persistence mechanisms (cron jobs, startup scripts, state files). Skills should not maintain state across sessions without explicit user consent.
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 | $0.00037 | $0.01199 |
| Opus 5 | $0.00018 | $0.00600 |
| Sonnet 5 | $0.00007 | $0.00240 |
| Haiku 4.5 | $0.00004 | $0.00120 |
Grade A, and why
orchestrating-fusion 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 13d 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 — 86 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Fusion Orchestration
Separate judgment from mechanical implementation. A lead investigates, decides, specifies, reviews, and verifies. An executor edits only the bounded scope in the lead's specification.
Route Deliberately
Do not force every task through Fusion:
- Keep a tiny, already-understood, single-file mechanical change in the normal writable session.
- Use Fusion when investigation, architectural judgment, multiple non-trivial writes, or independent review materially improve the result.
- Offer a durable PRD, ADR, or implementation plan only when persistent artifacts reduce substantial ambiguity. Never create a spec lifecycle merely because a change is large or risky.
Native Adapters
Prefer the installed native roles when the current tool exposes them:
- OpenCode: run the
fusion-leadprimary; it can delegate tofusion-executorand cannot edit or run shell commands. - Amp: select the
fusionagent mode; its restricted tool surface exposesfusion_executorfor implementation. - Codex: from the writable root session, spawn
fusion-leadfor the specification, then spawn the siblingfusion-executor; a read-only child cannot safely escalate a nested child to workspace-write. - Pi: from the root session, use the
Agenttool to runfusion-lead, then run the siblingfusion-executor;pi-subagentsenforces the lead's tool allowlist and intentionally disables nested delegation.
For other assistants, apply the workflow below using their native subagent/task tool. Treat role separation as advisory unless the harness actually restricts each role's tools or sandbox.
Workflow
- Have the lead inspect the request and relevant code. It may delegate read-only discovery.
- The lead resolves architectural and product choices. Ask the user only when a missing decision materially changes the result.
- The lead selects only the relevant installed skills. Include each exact
SKILL.mdpath in the handoff; do not paraphrase a skill into a lossy substitute or preload unrelated skills. - The lead sends the executor one bounded specification using this contract. Keep specifications concise (target under 2,000 tokens) — summarize objectives succinctly and reference skill paths, rather than pasting full PRDs, raw user prompts, or file contents verbatim. When the harness prevents nested delegation, the lead returns this specification to the writable root, which starts the executor as a sibling:
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.
- 13d ago First seen · 86 lines · 37 tokens per session scan A 6deb8d5396cc
orchestrating-fusion is a skill published in the GitHub repository jellydn/my-ai-tools (120 stars, last pushed today), licensed MIT. It adds 37 tokens to every session and 1,199 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 skills, from other repositories
geo-explorer
A Chinese-language skill that explores countries, regions, and international topics through geography, history, resources, and relationships between places.
strategic-product-advisor
A Chinese-language product-strategy guide for evaluating what a product is for, whom it serves, and how it could grow.
ask-first
Use only when explicitly invoked to clarify ambiguous, analogy-driven, meta-level, weakly delegated, or under-specified requests before execution.
insight-miner
Only invoke when explicitly requested via "洞见"、"@insight-miner" or "insight". Do NOT auto-trigger.
book-recommender
A Chinese-language book recommendation and analysis skill that evaluates what books are about, whether they are worth reading, and how to read them.
feynman-write
A research-and-writing workflow based on the Feynman method, where the AI gathers source material and the author decides the article's focus, then explains the ideas through questioning. The final text keeps the author's own examples and voice while adding technical precision.