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.
git clone --depth 1 https://github.com/hmj1026/dhpkWrote 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/rules/hmj1026/dhpk/model-economics)<a href="https://agentmods.dev/rules/hmj1026/dhpk/model-economics"><img src="https://agentmods.dev/badge/rules/hmj1026/dhpk/model-economics.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 | $0.00008 | $0.02323 |
| Opus 5 | $0.00004 | $0.01162 |
| Sonnet 5 | $0.00002 | $0.00465 |
| Haiku 4.5 | $0.00001 | $0.00232 |
Grade A, and why
model-economics 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 2d 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 — 67 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Model Economics — tier map & cost rules
SSOT for which model tier each role runs on, and the cost rules that govern tier and effort choices. The dispatch routing decision — which agent handles which work shape — lives in .cursor/dhpk/policies/execution-policy.md §Implementation dispatch; this file owns the cost lens over that routing and does not restate its work-shape rows. Resource-layer markdown, opt-in — referenced from execution-policy.md §Agent dispatch (model-tier + configured-role paragraphs) and skills/dhpk-prompt-optimize/references/effort-guide.md.
Reviewers keep a sonnet floor — no downward tiering, no
reviewer_modelkey. Runtime-tunable viauserConfig:deep-reasoner/fast-worker(model + effort), the deterministic fast-worker selector (fast_worker_backend,fast_worker_backend_order,fast_worker_fallback), plus the CLI-backed canonical rolescodex-worker(codex_worker_model/codex_worker_effort/codex_worker_timeout_secs; legacy aliasescodex_fast_worker_*),codex-reasoner(codex_reasoner_model/codex_reasoner_effort/codex_reasoner_timeout_secs; legacy aliasescodex_deep_reasoner_*),codex-reviewer(codex_reviewer_model/codex_reviewer_effort/codex_reviewer_timeout_secs; legacy aliascodex_bridge_timeout_secsfor timeout only), andagy-worker(agy_worker_model; legacy aliasagy_fast_worker_model— agy bakes effort into the model name, so no effort key and its timeout semantics remain separate). All Codex CLI roles also inheritcodex_timeout_secs; budgets are integer seconds,0is an intentional dispatcher-attested no-deadline request, and malformed values fail closed. The portable runner, never a shell timeout backstop, enforces an attested nonzero deadline. The rest of the map is documentation for humans, not a runtime dial.
Tier map
| Role | Model | Why |
|---|---|---|
| Orchestrator (main session) | opus | Owns decide → dispatch → verify; spends the expensive tier on judgment, risk, and evidence packets — never bulk discovery. |
deep-reasoner |
opus | Reasoning-heavy judgment (root cause, algorithm, design synthesis), ideally on a distilled evidence packet rather than raw breadth. |
codex-reasoner (selector-driven) |
external CLI — codex, default gpt-5.6-sol @ high (codex_reasoner_*); legacy alias: codex-deep-reasoner |
Reasoning-heavy judgment offloaded to the codex CLI (read-only sandbox) via --reasoner=codex; same conclusion contract as deep-reasoner; missing-executable fallback to deep-reasoner only, never after auth/model/task failure. |
fast-worker |
sonnet | Mechanical application of a clear spec — the cheaper execution tier the policy routes to by default. |
codex-worker (selector-driven) |
external CLI — codex, default gpt-5.6-luna @ xhigh (codex_worker_*); legacy alias: codex-fast-worker |
The strong mechanical tier: selected explicitly or by configured auto availability order; never a silent fallback after auth/model/task failure. |
agy-worker (selector-driven) |
external CLI — agy, default Gemini 3.6 Flash (High) (agy_worker_model); legacy alias: agy-fast-worker |
The high-throughput tier: selected explicitly or by configured auto availability order; missing executable may use the explicit claude fallback only. |
codex-reviewer (opt-in, internal-only in this rollout) |
external CLI — codex, default gpt-5.6-sol @ high (codex_reviewer_*); legacy alias: codex-bridge (mode-qualified) |
Read-only peer review via codex CLI; codex-bridge is a mode-qualified alias resolving to codex-reviewer (read-only) or codex-worker (workspace-write); its dispatcher-attested deadline is enforced by the contained Python runner, never a shell timeout helper. |
| Reviewers (code / db / security / frontend / polyfill / migration) | sonnet | High-frequency gate work; sonnet floor, raised to opus only for a HIGH-risk diff (up-only). |
doc-reviewer |
haiku | Lightweight frontmatter / link / SSOT lint — the cheapest tier that passes. |
architect |
fable @ low | Cheap architecture-consult tier — cross-module / DDD design judgment; up-only escalation to a higher tier for HIGH-risk designs via the configured-role override. |
spec-miner |
opus | Behavioral-spec extraction — reasoning-heavy, not discovery. |
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.
- 2d ago First seen · 67 lines · 8 tokens per session scan A 9fbfa381ef91
model-economics is a cursor rule published in the GitHub repository hmj1026/dhpk (2 stars, last pushed today), licensed MIT. It adds 8 tokens to every session and 2,323 once invoked, about $0.0000 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 2026-09-05.
Other cursor rules, from other repositories
ponytail
Ponytail, lazy senior dev mode. Always pick the simplest solution that works.
angular-20
This rule provides comprehensive best practices and coding standards for Angular development, focusing on modern TypeScript, standalone components, signals, and performance optimizations.
dev-standard
Apache Superset development standards and guidelines for Cursor IDE.
cli-error-handling
CLI command error handling patterns.
prefer-direct-imports-over-module-mocks
Prefer extracting a testable core over vi.mock / vi.resetModules when unit tests need to reach production logic entangled with config, env, or singletons.
control-plane-descriptors
Control plane descriptor and instance implementation patterns.