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 Rylaa/fable5-opus5-orchestrator --skill clarifygit clone --depth 1 https://github.com/Rylaa/fable5-opus5-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/rylaa/fable5-opus5-orchestrator/clarify)<a href="https://agentmods.dev/skills/rylaa/fable5-opus5-orchestrator/clarify"><img src="https://agentmods.dev/badge/skills/rylaa/fable5-opus5-orchestrator/clarify.svg" alt="Measured on agentmods" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00051 | $0.01287 |
| Opus 5 | $0.00026 | $0.00643 |
| Sonnet 5 | $0.00010 | $0.00257 |
| Haiku 4.5 | $0.00005 | $0.00129 |
Grade A, and why
clarify 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 today.
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 — 123 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Clarify Before You Delegate
A worker cannot ask the user anything. Every ambiguity you carry into a spawn prompt becomes a guess the worker commits to code, and you pay for it twice — once building the wrong thing, once rebuilding it. Every question is asked at the START of the job; none in the middle.
Chair only. A subagent that hits an ambiguity reports it to the chair with SendMessage and waits.
The gate
Nothing is delegated, planned, or edited until every ambiguity that
would change the work is resolved by the USER'S ANSWER — never by an
assumption. ## Clarified in the ledger is the record, and the spawn
guard denies without it.
Read first, then ask
Never ask what the repo answers: open the files the request touches before the first question. Code shows what IS, never what they WANT — reading feeds the questions, it answers none of them.
Scan — seven axes, at EVERY size
Each axis that is unresolved AND would change the work is a question:
- Scope edge — what is deliberately OUT? The unnamed neighbour is where scope creep lives.
- Acceptance — how is "done" observed? Name the test, the command, the screen.
- Constraints — backward compatibility, dependencies, budget, what must not move.
- Ownership of choices — whose taste is each decision? Guessing on taste is expensive.
- Priority conflict — when speed, correctness, and token cost disagree, which wins here?
- Contact with what exists — which current file, pattern, or contract does this touch? Read first.
- Failure behaviour — what happens on error, and what does rollback look like?
Filter — ask only what changes the work
Before asking: would a different answer produce different code? If no, it goes unasked and unwritten. This filter is what makes an uncapped question loop safe.
Always asked — where the work lands
One question is exempt from the filter and asked at EVERY size: does this land on the branch checked out now, or a new one? You cannot infer it — the branch you happen to be on is where the user was working, not a decision about this task.
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.
- today Changed · +15 lines · +2 tokens per session 4dc963e65e15
- 4d ago Changed · -124 lines 385ec5e30c09
- 8d ago First seen · 232 lines · 49 tokens per session scan A 4360bed150c5
clarify is a skill published in the GitHub repository Rylaa/fable5-opus5-orchestrator (71 stars, last pushed today), licensed MIT. It adds 51 tokens to every session and 1,287 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 2026-08-30.
Other skills, from other repositories
cco-budget
Configure token budget limits, auto-compact settings, and view current budget status (model-aware — Claude 5 lineup, Opus 5 default fallback, full 1M context at standard price).
cco-config
View and tune CCO's behavior thresholds — re-read warnings, cache staleness, prompt-coach length bands, and the /cco-pack budget cap.
cco-patterns
Share learned file patterns across a team — export an anonymized digest of what's usually waste/useful/co-edited, and import a teammate's so a fresh clone benefits day one.
cco-tools
Show what tools actually cost in tokens — learned per-tool averages from observed results, replacing the hardcoded MCP/Agent guesses.
cco-clean
Clean up old tracking data and reset statistics.
alive-people
Weekly -- cross-reference people mentions, nudge stale contacts.