Skills Data Science Analyze Session Message Costs

Analyze Session Message Costs

v20260707
cost-session
This tool provides a granular, per-message cost breakdown within a single conversation session. It serves as a deep-dive utility following an anomaly detection, allowing operators to pinpoint exactly which messages incurred the highest costs. It reveals critical cost components like output tokens, cache writes, and model escalations, helping users understand the true resource consumption beyond simple token counts.
Get Skill
202 downloads
Overview

When cost-anomaly flags a session as a >3.5σ outlier, the next question is "which MESSAGES were expensive?". cost-session answers that.

Question Skill
"Which sessions cost the most?" cost-conversation
"Which sessions are outliers?" cost-anomaly
"Which messages in THIS session were expensive?" cost-session ← this

Algorithm

Implementation: scripts/session.mjs.

  1. Resolve session jsonl: --session-id <id> (scans ~/.claude/projects/*/) or --latest (default; picks most-recently-modified jsonl).
  2. Parse all assistant messages with usage blocks.
  3. Cost each message via shared PRICING (_prices.mjs).
  4. Sort descending by cost_usd, surface top-N (default 20).
  5. Compute p50/p90/p99 of message costs for in-session percentile context.
  6. Flag the top message if it's >2× the p99 — that's an in-session outlier.

Why this matters: cache writes are the silent cost

Example real session, top message:

| # | Model    | In | Out | Cache W | Cache R | Cost      |
| 1 | opus-4-7 | 6  | 569 | 881898  | 0       | $16.58    |

Without the Cache W column it looks like "569 output tokens cost $16" — that's wrong by 380×. The actual cost is ephemeral 1h cache write at opus pricing: 881,898 tokens × $18.75/1M = $16.54.

Operators reading the table see immediately: "the model wrote 881K tokens to ephemeral cache". From there the question becomes "why did we cache 881K tokens of context for a 6-input request?" — that's a real engineering signal.

Drill-down workflow

# Step 1: find outliers across all sessions
cost anomaly --alert-on-outliers 1 || cost anomaly  # see which session-ids

# Step 2: drill into the flagged session
cost session --session-id <flagged-id> --top 10

# Step 3: open that jsonl at the timestamp the top message reports,
# inspect the prompt + tool calls

Percentile context

Top of output:

| p50 (median) message | $0.85 |
| p90 message          | $1.45 |
| p99 message          | $1.74 |

Lets operators ask "is this top message a 2× outlier or a 380× one?" without having to compute it themselves. The "top is >2× p99" footer fires when the answer is "yes, this is an in-session outlier worth investigating".

--since filter

Useful for drilling into a specific time range within a long session:

cost session --since 2026-06-16T13:00:00Z --top 5

Only messages with timestamp >= --since are considered.

Edge cases

  • No costed assistant messages → "No costed assistant messages" + exit 0.
  • --session-id not found in any project's jsonls → exit 2 with error.
  • --top must be a positive integer → exit 2.
Info
Category Data Science
Name cost-session
Version v20260707
Size 3.07KB
Updated At 2026-07-09
Language