Works with the AI Agents you already use
Agent Cost Profiler
Find where an agent's token spend actually goes, separate waste from necessary cost, and cut it without breaking quality.
$19
Agent Cost Profiler
Example session with this skill installed
Our LLM bill went from $800 to $6,400 a month and nobody knows why. Here is the usage export: task, model, input tokens, output tokens, retries. Build me a cost model and tell me what to cut.
- Read your context and instructions
- Compiled the agent cost profiler
Cost profile: support agent
Headline
Current: $6,400/month across 240k requests. Cost per successful task: $0.027.
Largest driver: conversation history at 41% of spend.
Realistic saving: $4,900/month (77%), of which $3,400 is low risk.
Attribution
| Layer | $/mo | Share | Necessary? |
|---|---|---|---|
| Conversation history | 2,620 | 41% | Reducible |
| Classification calls | 1,450 | 23% | Reducible |
| Tool payloads | 1,180 | 18% | Reducible |
| Retry storms | 640 | 10% | Waste |
| Reasoning step | 510 | 8% | Necessary |
Findings
F1 — History is never trimmed
Evidence: traces t-1180, t-2044 — 14 turns, 22k input tokens, last 12 unused.
Cost: $2,620/month.
Fix: summarise beyond four turns, keep the last two verbatim. Risk: low. Verify: 50 multi-turn eval cases.
F2 — Retry storm on a deterministic error
Evidence: 18% of lookup_order calls return the same 404 three times.
Cost: $640/month. Pure waste.
Fix: stop on deterministic 4xx and surface the error.
F3 — Frontier model on classification
Evidence: 60% of requests are intent labels; a small model matched 40/40 in a parity check.
Cost: $1,450/month. Saving: $1,180 with low risk.
Ranked interventions
| # | Intervention | Saving | Effort | Risk |
|---|---|---|---|---|
| 1 | Cap and summarise history | $1,900 | M | Low |
| 2 | Route classification to a small model | $1,180 | M | Low |
| 3 | Filter tool payloads to needed fields | $880 | S | Low |
| 4 | Stop deterministic retries | $640 | S | Low |
| 5 | Cache the system prompt | $310 | S | Low |
What not to cut
The reasoning step that produces refund decisions. It is 8% of spend and the source of the product's quality.
Assumptions
Rates from the supplied export. Volume assumed flat; at 2x volume the history fix saves $3,800.
Connects securely to your tools. The creator never sees your data.
About this skill

Requires ingestion of usage logs or billing exports. Cannot connect directly to cloud providers. Decisions rely on user-provided pricing rates.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 2 days ago