Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/kimsanguine/hplannpx agentmods add skills/kimsanguine/hplan/cogs-sentinelWrote 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/kimsanguine/hplan/cogs-sentinel)<a href="https://agentmods.dev/skills/kimsanguine/hplan/cogs-sentinel"><img src="https://agentmods.dev/badge/skills/kimsanguine/hplan/cogs-sentinel.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 | $0.00103 | $0.01306 |
| Opus 5 | $0.00051 | $0.00653 |
| Sonnet 5 | $0.00021 | $0.00261 |
| Haiku 4.5 | $0.00010 | $0.00131 |
Grade A, and why
cogs-sentinel 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 7d 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 — 99 lines — stays where its author put it; the contents beside it link to each section on GitHub.
COGS Sentinel — Executable Margin Gate
Running for: $ARGUMENTS
Core Goal
- Token 단가와 사용 패턴으로부터 p50/p90 월간 COGS와 gross margin을 계산하여 paid AI product의 경제적 viability를 PRD 전에 판단한다.
- Free-user abuse 시나리오를 blend해 "유료 1명당 무료 24명을 감당해야 할 때" 마진이 살아남는지 본다.
- 결정은 GREEN / CONDITIONAL_GO / RED + 구체적 mitigation 권고.
Trigger Gate
Use This Skill When
- 사용자가 유료 가격을 책정하려 할 때 (
$X/mo,per-call, etc.) discover/cost-sim이 시나리오를 도출한 후 결정론적 수치가 필요할 때- Provider 교체(Anthropic ↔ OpenAI ↔ Google)의 영향 측정
- 무료 tier 도입 시 abuse가 마진을 깨지 않는지 사전 검증
- Build Gate 진입 전
Route to Other Skills When
- 비용 시나리오 자체가 아직 명확하지 않을 때 →
discover/cost-sim먼저 - 배포 후 실제 추적 →
operate/ops-review - 가격 모델 자체를 다시 짤 때 →
architect/strategy - COGS RED 결정이 났을 때 →
decision-log(hold/pivot) 기록 후 routing
Boundary Checks
- ℹ️
--cac와--monthly-churn을 함께 전달하면 LTV/CAC/Payback + overall_verdict(BUILD/HOLD/INVESTIGATE)를 추가로 산출한다. 두 인자를 생략하면 무시되어 하위호환이 유지된다. (--mau추가 시 Business Report 블록 — 매출/COGS/cost_ratio/blended — 도 포함됨.) - ℹ️ 이 skill은 sanity-check(빌드 게이트)이지 회계 보고서가 아니다 — 법인세·감가상각 등 전산 회계 항목은 다루지 않는다.
- ❌ Provider 단가는
references/provider_pricing.json스냅샷 기준이다. 최신 단가는 사용자가 verify해야 함. - ❌ lognormal sampling이라 결정론적이지만 실제 분포와 다를 수 있다 — sanity check 용도지 회계 보고서 아님.
Inputs
CLI:
python3 hplan/scripts/cogs_sentinel.py \
--provider anthropic --model claude-sonnet-4-6 \
--tokens-in 3000 --tokens-out 1500 --calls-per-user-month 40 \
--arpu 29 --paid-conversion 0.08 --free-abuse-multiplier 3
또는 JSON:
{
"provider": "anthropic", "model": "claude-sonnet-4-6",
"tokens_in": 3000, "tokens_out": 1500,
"calls_per_user_month": 40, "arpu": 29,
"paid_conversion": 0.08, "free_abuse_multiplier": 3,
"target_gross_margin": 0.70
}
Steps
- Confirm provider + model exists in
references/provider_pricing.json(or pass--pricing path). - Estimate usage params from discover/cost-sim output or user input.
- Run
python3 hplan/scripts/cogs_sentinel.py .... - Read the markdown report at
harness/build-gate/cogs_report.md. - If decision is
GREEN, proceed todecision-log→handoff. - If
CONDITIONAL_GO, list mitigations and require human approval. - If
RED, route todecision-logwith decision =pivotorhold.
What ships with it
4 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.
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.
- 7d ago First seen · 99 lines · 103 tokens per session scan A 03812880ee7f
cogs-sentinel is a skill published in the GitHub repository kimsanguine/hplan (2 stars, last pushed 22d ago), licensed MIT. It adds 103 tokens to every session and 1,306 once invoked, about $0.0005 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
invest
First-time fork users: run the invest-setup skill first to initialize. The main flow lives in this skill (the AI agent uses the CLI/MCP to view the portfolio / run the committee / replay decision history). The Web GUI has been retired (2026-07) — every capability is exposed via CLI subcommands / MCP tools. The backend…
invest-setup
A first-time setup workflow for openInvest, a system for tracking investments and running portfolio analysis.
invest-backup
Back up / restore openInvest's local state — memory/ (holdings, strategy, user profile, committee records, dream logs) + db/ (trade ledger, job run history, market-data cache) + .env (SMTP/API credentials) + userprofile.json. All of this data is .gitignore'd with no historical versions in git, so a single accidental…
portfolio
Portfolio system knowledge base. Use when: position queries, routing strategy questions, provider integration. Not for: general code exploration (use code-explore), code review (use codex-code-review). Output: domain-specific analysis + recommendations.
yield-xyz-agentkit-builder
Build applications that integrate with the Yield.xyz APIs — staking, lending, vaults, and RWA across 80+ networks. Generates code covering REST API integration, transaction signing, wallet connection, and fee monetization. Use when user wants to build an app, integrate yield, generate code, or set up a project using…
yield-xyz-agentkit-privy
The Privy connector for the Yield.xyz AgentKit — signs and broadcasts via Privy agentic wallets. Extends the yield-xyz-agentkit skill — that skill discovers yields and builds the unsigned transactions; this one adds Privy wallet creation, policy enforcement, signing, and broadcasting, in autonomous and semi-autonomous…