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 orenluxy/fable-method --skill model-economygit clone --depth 1 https://github.com/orenluxy/fable-methodWrote 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/orenluxy/fable-method/model-economy)<a href="https://agentmods.dev/skills/orenluxy/fable-method/model-economy"><img src="https://agentmods.dev/badge/skills/orenluxy/fable-method/model-economy.svg" alt="Measured on agentmods" 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 | $0.00070 | $0.00544 |
| Opus 5 | $0.00035 | $0.00272 |
| Sonnet 5 | $0.00014 | $0.00109 |
| Haiku 4.5 | $0.00007 | $0.00054 |
Grade A, and why
model-economy 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 6d 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.
Model Economy
Principle: a standing instruction like "use the cheapest model" does nothing by itself — the main session model cannot swap itself. The ONLY mechanism for per-step model choice is delegation to subagents with an explicit model: field. Cost savings happen when mechanical work is pushed down and judgment work stays up.
Delegation table
| Step type | Route to | Why |
|---|---|---|
| Territory scanning: codebase exploration, convention extraction, web fact-finding | territory-scout (haiku) |
Wide, read-only, mechanical. Findings > prose. |
| Change inventory, diff summarization, test-run reporting | report-builder (sonnet) |
Mechanical but needs accurate code reading. |
| Bulk transformations with a clear spec (renames, format conversions, boilerplate) | subagent on haiku/sonnet, spec written by main model | Spec quality is the judgment; execution is mechanical. |
| Architecture decisions, interviews, quiz design, deviation calls, anything user-facing | main model, no delegation | Judgment and full context required. Delegating this is false economy. |
| Prototype variations (prototype-first) | main model | Taste extraction IS the point; cheap models produce generic variations that waste the pass. |
Rules
- Never delegate judgment to save tokens. A wrong architectural call costs more than every scan in the project combined. Cheap models get facts-gathering, not decisions.
- Write the brief before delegating. A subagent with a vague brief returns noise you'll re-do on the expensive model — negative savings. One paragraph: what to look for, where, output format.
- One round-trip. If a subagent's output needs a second clarification round, absorb the task into the main model instead — the coordination overhead has eaten the savings.
- Report the routing. When you delegate, tell the user in one line: "scanning via territory-scout (haiku) to keep main context lean." This keeps the economy auditable.
- If the environment has no subagent support (e.g., claude.ai), state that model routing is unavailable and proceed on the current model — do not pretend to route.
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.
- 6d ago First seen · 27 lines · 70 tokens per session scan A 7471af1f3db5
model-economy is a skill published in the GitHub repository orenluxy/fable-method (1 stars, last pushed 1mo ago), licensed MIT. It adds 70 tokens to every session and 544 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 skills, from other repositories
journey-simulation
Use when caller wants to observe how a stranger encounters a flow, artifact, or sandbox — triggers like "simulate a user journey", "test our onboarding / checkout / signup", "will my ICP convert", "how does a cold reader experience this README", "first-time user test", "cognitive walkthrough", or any request to…
release-init
Detect project type and generate a tailored project-level /release skill. Run once per project to set up releasing.
answer-all-open-questions-with-recommendation
Record the embedded recommendation as the answer for every open and deferred question in the current milestone's requirements that carries one.
derive-tasks
Convert the current milestone's requirements.md into a complete, dependency-ordered TASKSTODO.md of brief-level task sections.
answer-open-question-with-recommendation
Record an open question's embedded recommendation as that question's answer in the current milestone's requirements.md.
discuss-open-question
Discuss a named open question in the current milestone requirements, surfacing alternatives, trade-offs, and a recommendation to help the user decide.