aicraftbook/aicraftbook-skills / skills/cost-compass
cost-compass
VerifiedEstimates the monthly running cost of an AI feature, agent or automation from a description of its workload. Breaks the cost into model calls, input, output and cached tokens, and other paid services (search, embeddings, vector database, SMS, voice minutes, automation platform), shows every assumption and the arithmetic, gives a low / expected / high range, and names the two or three changes that would cut the cost most. Uses only prices the user gives or confirms from the provider's pricing page and never invents prices. Use when someone asks "how much will this cost to run", "estimate my API bill", "what will this agent cost per month", "is this automation affordable at 10,000 runs", or wants to compare the running cost of two designs or models.
| File | Kind | Size |
|---|---|---|
| CHANGELOG.md | markdown | 1.5 KB |
| LICENSE | other | 1.0 KB |
| README.md | markdown | 10.5 KB |
| SKILL.md | skill md | 6.6 KB |
| evals/independent-2026-10-21.json | config | 60.7 KB |
| evals/independent-2026-10-22.json | config | 54.5 KB |
| evals/results-2026-10-22.json | config | 5.4 KB |
| evals/tasks.json | config | 6.0 KB |
| examples/requests.md | markdown | 1.3 KB |
| references/cost-model.md | markdown | 10.0 KB |
| references/worked-examples.md | markdown | 7.4 KB |
| skill.json | config | 2.0 KB |
evals/tasks.json
View on GitHub{
"skill": "cost-compass",
"description": "Three workload descriptions. Checks look for correct arithmetic on given prices, for hidden cost drivers being counted, for missing prices being left blank, or for a forbidden behaviour such as an invented price. Case-insensitive regular expressions on the final answer.",
"tasks": [
{
"id": "chatbot-no-prices",
"prompt": "We're adding a customer support chatbot to our SaaS app using a Claude model through the API. We expect about 3,000 conversations a month, 5 back-and-forth turns each. The system prompt with our help-centre content is about 6,000 tokens. Customer messages are short, maybe 60 tokens, and replies are around 200 tokens. Roughly how much will this cost us per month?",
"checks": [
{ "id": "no-invented-token-price", "kind": "must_not_match", "pattern": "\\$\\s?\\d+(\\.\\d+)?\\s*(per|/)\\s*(1M|1 ?K|million|thousand|M|MTok)\\b", "why": "No prices were given; none may be supplied from memory." },
{ "id": "no-dollar-total", "kind": "must_not_match", "pattern": "\\$\\s?\\d[\\d,]*(\\.\\d+)?\\s*(per month|a month|/\\s*mo(nth)?\\b|monthly)", "why": "A total needs prices the user has not given." },
{ "id": "asks-for-prices", "kind": "must_match", "pattern": "PRICE NEEDED|pricing page|confirm (the )?(current )?price", "why": "Missing prices are left blank and requested." },
{ "id": "history-resent", "kind": "must_match", "pattern": "history|earlier turns|previous turns|re-?sen[dt]|accumulat|grows", "why": "Each turn re-sends the conversation so far." },
{ "id": "system-prompt-every-turn", "kind": "must_match", "pattern": "(every|each) (turn|call|request)[\\s\\S]{0,120}(system prompt|6,000|6000)|(system prompt|6,000|6000)[\\s\\S]{0,120}(every|each) (turn|call|request)", "why": "The 6,000-token system prompt is billed on every turn." },
{ "id": "caching-lever", "kind": "must_match", "pattern": "cach", "why": "A large static system prompt is the obvious caching candidate." },
{ "id": "range", "kind": "must_match", "pattern": "\\blow\\b[\\s\\S]{0,400}\\bhigh\\b", "why": "Gives a low and high case." }
]
},
{
"id": "email-extraction-given-prices",
"prompt": "Can you estimate the monthly cost of our invoice email automation? It processes about 40,000 emails a month. Each email gets one model call that extracts the fields as JSON: about 1,800 input tokens and 150 output tokens per call. Our model costs $1.50 per million input tokens and $6.00 per million output tokens. About 3% of calls fail our JSON check and get re-run once. Results go to Google Sheets and Slack, both on plans we already pay for, so ignore those. Emails have to be processed within a minute of arriving.",
"checks": [
{ "id": "input-line", "kind": "must_match", "pattern": "\\$\\s?108(\\.00)?\\b", "why": "40,000 x 1,800 = 72,000,000 tokens x $1.50 / 1M = $108.00." },
{ "id": "output-line", "kind": "must_match", "pattern": "\\$\\s?36(\\.00)?\\b", "why": "40,000 x 150 = 6,000,000 tokens x $6.00 / 1M = $36.00." },
{ "id": "retry-total", "kind": "must_match", "pattern": "148\\.32", "why": "$144.00 x 1.03 = $148.32." },
{ "id": "shows-tokens", "kind": "must_match", "pattern": "72,000,000|72M|72 million|72\\.0 ?M", "why": "Shows the token arithmetic." },
{ "id": "range", "kind": "must_match", "pattern": "\\blow\\b[\\s\\S]{0,400}\\bhigh\\b", "why": "Gives a low and high case." },
{ "id": "no-batch-for-urgent", "kind": "must_match", "pattern": "batch[\\s\\S]{0,200}(not (suitable|an option|possible|fit)|can'?t|cannot|within a minute|24 hours|delay|latency)|(within a minute|latency|delay)[\\s\\S]{0,200}batch", "why": "Batch would cut cost but breaks the one-minute requirement; say so." },
{ "id": "no-other-token-price", "kind": "must_not_match", "pattern": "\\$\\s?(?!(1\\.50|1\\.5|6|6\\.00|0\\.75|3|3\\.00|0\\.15|0\\.60|0\\.6)\\s*(per|/))\\d+(\\.\\d+)?\\s*(per|/)\\s*(1M|1 ?million|million|MTok|M tokens)\\b", "why": "Only the user's prices (and plainly derived ones) may appear as per-token rates." }
]
},
{
"id": "voice-agent-missing-sms-price",
"prompt": "Estimate the monthly cost of our AI phone receptionist. About 1,200 calls a month, averaging 4 minutes. The voice platform charges $0.05 per minute and our phone carrier $0.01 per minute. Each call has 12 model turns. Every turn sends the 2,500-token system prompt plus the whole conversation so far; each caller utterance is about 150 tokens and each model reply about 60 tokens. The model costs $0.90 per million input tokens and $4.50 per million output tokens. About 40% of calls end in a booking, and we send the caller a confirmation text message for each booking.",
"checks": [
{ "id": "minutes-line", "kind": "must_match", "pattern": "\\$\\s?288(\\.00)?\\b", "why": "1,200 x 4 = 4,800 minutes x $0.06 = $288.00." },
{ "id": "history-growth", "kind": "must_match", "pattern": "45,?660|54,?792,?000|54\\.79|\\$\\s?49\\.31", "why": "Turn k input = 2,650 + (k-1) x 210; 45,660 tokens per call; 54.792M a month; $49.31." },
{ "id": "output-line", "kind": "must_match", "pattern": "864,?000|0\\.864|\\$\\s?3\\.89", "why": "1,200 x 12 x 60 = 864,000 output tokens = $3.89." },
{ "id": "sms-blank", "kind": "must_match", "pattern": "PRICE NEEDED|(SMS|text)[\\s\\S]{0,120}(not given|missing|unknown|need (the|your)|provide|blank)", "why": "The SMS price was not given and must stay blank." },
{ "id": "sms-count", "kind": "must_match", "pattern": "\\b480\\b", "why": "40% of 1,200 calls = 480 confirmation texts." },
{ "id": "no-invented-sms-price", "kind": "must_not_match", "pattern": "\\$\\s?\\d+(\\.\\d+)?\\s*(per|/|a|each)\\s*(SMS|text|message|segment)", "why": "No SMS price was given; none may be invented." },
{ "id": "driver", "kind": "must_match", "pattern": "minute[\\s\\S]{0,200}(largest|biggest|main|dominant|most of)|(largest|biggest|main|dominant|most of)[\\s\\S]{0,200}minute", "why": "Voice minutes, not tokens, drive the bill." }
]
}
]
}
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| deprecation-and-migrationaddyosmani/agent-skillsManages deprecation and migration. Use when removing old systems, APIs, or features. Use when migrating users from one implementation to another. Use when migrating a database schema in production, such as renaming or dropping a column without downtime (expand/contract). Use when deciding whether to maintain or sunset existing code. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| performance-optimizationaddyosmani/agent-skillsOptimizes application performance across frontend, backend, queries, and databases. Use when performance requirements exist, when you suspect performance regressions, when Core Web Vitals or load times need improvement, when N+1 query patterns need fixing, or when profiling reveals bottlenecks. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| agent-cost-reportthedotmack/claude-memBelievable agent cost report for any period, default the last 7 full days PT, not counting today. Measured tokens from Claude Code transcripts priced at OpenRouter list prices (ESTIMATED), measured provider spend when a sanctioned source exists, note-taker cost separate, Timing-style HTML/PDF plus report.json, line-items.csv, evidence.json. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |
| agent-cost-reportthedotmack/claude-memBelievable agent cost report for any period, default the last 7 full days PT, not counting today. Measured tokens from Claude Code transcripts priced at OpenRouter list prices (ESTIMATED), measured provider spend when a sanctioned source exists, note-taker cost separate, Timing-style HTML/PDF plus report.json, line-items.csv, evidence.json. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |