Indigo

Engineer · AI usage, harness practices and where spend could have gone further.

Demonstration report. This engineer and their usage are synthetic.

This page is for the engineer and their manager to find cheaper defaults and harness fixes. It is never used to rank people.

Period

Summary

FY2027 to date

AI spend

$1,798

187 sessions · $10 each

Savings opportunity

$456

25% of spend, at list price

Practices on target

2 of 7

Focus: Knowledge MCP use, rtk coverage

Cache hit rate

82%

Team 82% · org 82%

FY2027 to date · estimated at list price

$456 of $1,798 spend

Largest saving first

Improve prompt caching$259
Cache hit rate 82% against an 85% target.
Keep stable context at the front of prompts and avoid switching models mid-session, so 85% of input reads from cache.
Right-size models$120
46% of routine spend ran on Opus: code research $163 of $342, documentation $60 of $147, ops $47 of $99, presentation $24 of $57.
Run docs, decks, ops and read-only research on Sonnet instead of Opus.
Keep context lean$57
10% of input tokens were past 100K of context.
Compact or /clear between tasks instead of carrying more than 100K tokens of context.
Route commands through rtk$20
61% of 1,716 Bash calls ran through rtk; 2.8M tokens of command output went into context.
rtk compacts test, build, git and search output before it reaches the model (about 70% smaller).

Compared with Social Publishing and the organization

Practice	Indigo	Team	Org	Target	Recommendation
rtk coverage	61%Needs attention	63%	56%	≥ 80% of Bash calls	Install the rtk hook so test, build, git and search commands return compact output.
Knowledge MCP use	3%Off target	4%	30%	≥ 40% of sessions	Search the knowledge MCP before re-researching a codebase, and save findings back with remember.
Cache hit rate	82%Needs attention	82%	82%	≥ 85% of input	Keep CLAUDE.md and tool lists stable, avoid mid-session model switches, and resume sessions instead of restarting.
Context past 100K	10%On target	9%	10%	≤ 10% of input	On target
Opus on routine work	46%Needs attention	44%	44%	≤ 25% of routine spend	Default routine sessions to Sonnet and keep Opus for harness, design and hard debugging work.
Sessions using subagents	47%On target	46%	50%	≥ 30% of sessions	On target
Tool error rate	4.1%Needs attention	3.7%	3.0%	≤ 4% of tool calls	Fix recurring failures (permissions, env vars, flaky tests) in the harness rather than letting the agent retry.

Is the engineer moving toward the targets?

Month	rtk	Knowledge	Cache	Context	Model fit	Subagents	Savings opportunity
Feb 2026	–	–	62%Off target	7%On target	62%Off target	38%On target	$65
Mar 2026	–	–	65%Off target	7%On target	62%Off target	25%Needs attention	$46
Apr 2026	–	–	67%Off target	6%On target	62%Off target	30%On target	$58
May 2026	18%Off target	–	72%Needs attention	9%On target	55%Off target	32%On target	$87
Jun 2026	43%Needs attention	0%Off target	76%Needs attention	7%On target	53%Off target	37%On target	$46
Jul 2026	65%Needs attention	5%Off target	80%Needs attention	8%On target	40%Needs attention	45%On target	$36
Aug 2026	69%Needs attention	4%Off target	79%Needs attention	10%Needs attention	38%Needs attention	49%On target	$69
Sep 2026	71%Needs attention	3%Off target	87%On target	11%Needs attention	30%Needs attention	58%On target	$29
Oct 2026	65%Needs attention	2%Off target	90%On target	11%Needs attention	27%Needs attention	53%On target	$19

FY2027 to date

By model tier, by month

Routine types (documentation, presentation, ops, code research) rarely need Opus

Cache reads are billed at a fraction of uncached input; a high share is what good caching looks like

Model	Requests	Uncached input	Cache reads	Cache writes	Output	Spend
Opus 5.5	2,579	8.1M	142.7M	9.6M	3.3M	$174
Opus 5	2,884	19.8M	139.3M	10.2M	4.1M	$335
Opus 4.8	1,724	15.1M	72.6M	5.6M	2.1M	$200
Opus 4.6	2,472	38.9M	101.2M	8.9M	3.8M	$397
Opus 4.5	168	3.3M	6.4M	619K	266K	$30
Sonnet 5.5	3,143	8.7M	161M	10.8M	4.6M	$106
Sonnet 5	3,959	26.2M	199.6M	14.4M	6.1M	$285
Sonnet 4.6	1,852	25M	78.7M	6.6M	2.7M	$164
Sonnet 4.5	161	3.2M	6.2M	602K	279K	$18
Haiku 5.5	3,471	9.2M	183.9M	12.3M	5.4M	$7
Haiku 4.5	2,751	24.5M	128.5M	9.8M	4.1M	$70
GPT Realtime (meetings)	157	1.1M	7.2M	528K	206K	$13
Total	25,321	183.1M	1227.3M	90M	36.9M	$1,798

Spend and tokens come from LiteLLM gateway logs and imported agent session history, priced at API list rates.
Savings are estimates: routine Opus work repriced at the Sonnet available at the time; uncached input moved to cache reads up to an 85% hit rate; rtk removing 70% of unfiltered command output, which stays in context for about half the session's remaining turns; and 50% of input past 100K context avoided by compacting. Levers overlap, so the total is an upper bound.
rtk and the knowledge MCP are scored from their rollout (rtk May 2026, knowledge MCP Jun 2026).
Generated Oct 20, 2026, 7:46 AM.