Findings
Rules over the same data: each finding carries a number and a tip.
Findings
examples/findings.ansiagentmeter · Claude Code · last 7 days
READ
✓ logs 1,299 files, 856,227 lines from ~/.claude/projects · version
2.1.170 – 2.1.309
✓ observed 2026-06-14T17:33:36Z – 2026-10-30T19:35:35Z · 130,425 requests in all
✓ quality every line readable
✓ rules 5 of 5 measurable in this source
✓ probe /context in 4 of 4 places (empty directory, web-casoon-dev, kestrel,
chartlet); answered locally, no model request · claude -p /context
--no-session-persistence --output-format json
FINDINGS ? 3 to look at ✓ 1 fine ✓ 1 nothing found
────────────────────────────────────────────────────────────────────────────────────────
? baseline-load look at it · notice · observed · v1
⇒ today's configuration is 34k of the median start 79k (43 %); 34k more is in every
start (front end, project instructions, a short first prompt) and 11k varies per
session; 146 of 146 starts came from claude-desktop, whose own instructions and
tools /context in the CLI does not see; project kestrel adds 8k (project
instructions or MCP servers)
spread of session starts ├──┃━━───────┤
smallest 68k · quartiles 76k / 87k · median 79k · largest 114k
median start, split ██████████████▓▓▓▓▓▓▓▓▓▓▓▓▓▓░░░░ 79k
█ configuration (/context, empty directory) 34k
▓ rest of the smallest start 34k
░ varies per session (median minus smallest) 11k
today's configuration, /context in an empty directory · claude-opus-5[1m]
System tools ████████████████████████ 19k
Skills ████████████▌··········· 10k
System prompt ███▊···················· 3k
Memory files █▊······················ 1k
Custom agents ▋······················· 514
loaded on demand, not sent at start: System tools (deferred) 29k · MCP tools
(deferred) 12k
MCP servers: webspire 9k · context7 3k · plugin_cloudflare_cloudflare-docs 561
memory files: ~/.claude/CLAUDE.md 1k
in the project directory: web-casoon-dev +4k · kestrel +8k · chartlet +400
? cache-rebuild look at it · notice · observed · v1
⇒ 202 rebuilds (60.1 M) are in sub-agent files, which wrote 100 % of their caches with
5 min while the main conversation wrote 100 % with 1 h; the log does not show
whether a setting reaches sub-agents; 147 of 349 unexpected rebuilds (42.4 M) follow
pauses over 1 h, which no cache lifetime covers
pause over 1 h █████████████████······· 42.4 M 147 rebuilds
5 min expired, main session ························ 0 0 rebuilds
5 min expired, sub-agents ████████████████████████ 60.1 M 202 rebuilds
cause unknown ························ 0 0 rebuilds
expected (model, compaction) ▌······················· 1.2 M 3 rebuilds
? long-session look at it · info · observed · v1
⇒ 22 sessions ran over a day, idle time included; only a change of topic inside one
makes a fresh session worth it, and the log does not show topics
runemark · 0f903e83 ████████████████████████ 155 h 500 requests
auditmysite · 1dadfa98 ██████████████████▏····· 117 h 116 requests
astro-site-files · 0f91e74f ██████████████████······ 116 h 389 requests
✓ big-context fine · notice · observed · v2
⇒ cheap while cached: 98 % of these tokens were cache reads; 39 % of these tokens sit
in 3 of 22 sessions; whether that length helped the task the log cannot show
kestrel · 21a6b55d ██████▌················· 360 M 509 requests
web-casoon · 2e02dbf0 ████▌··················· 250 M 355 requests
kestrel · 3ad3ed6b ████▏··················· 227 M 316 requests
19 other sessions ████████████████████████ 1.32 G 2,076 requests
✓ compact-thrash nothing to report: 0 of 15 automatic compaction pairs came with under
100k of context growth (v1)agentmeter · Claude Code · last 7 days
READ
✓ logs 1,299 files, 856,227 lines from ~/.claude/projects · version
2.1.170 – 2.1.309
✓ observed 2026-06-14T17:33:36Z – 2026-10-30T19:35:35Z · 130,425 requests in all
✓ quality every line readable
✓ rules 5 of 5 measurable in this source
✓ probe /context in 4 of 4 places (empty directory, web-casoon-dev, kestrel,
chartlet); answered locally, no model request · claude -p /context
--no-session-persistence --output-format json
FINDINGS ? 3 to look at ✓ 1 fine ✓ 1 nothing found
────────────────────────────────────────────────────────────────────────────────────────
? baseline-load look at it · notice · observed · v1
⇒ today's configuration is 34k of the median start 79k (43 %); 34k more is in every
start (front end, project instructions, a short first prompt) and 11k varies per
session; 146 of 146 starts came from claude-desktop, whose own instructions and
tools /context in the CLI does not see; project kestrel adds 8k (project
instructions or MCP servers)
spread of session starts ├──┃━━───────┤
smallest 68k · quartiles 76k / 87k · median 79k · largest 114k
median start, split ██████████████▓▓▓▓▓▓▓▓▓▓▓▓▓▓░░░░ 79k
█ configuration (/context, empty directory) 34k
▓ rest of the smallest start 34k
░ varies per session (median minus smallest) 11k
today's configuration, /context in an empty directory · claude-opus-5[1m]
System tools ████████████████████████ 19k
Skills ████████████▌··········· 10k
System prompt ███▊···················· 3k
Memory files █▊······················ 1k
Custom agents ▋······················· 514
loaded on demand, not sent at start: System tools (deferred) 29k · MCP tools
(deferred) 12k
MCP servers: webspire 9k · context7 3k · plugin_cloudflare_cloudflare-docs 561
memory files: ~/.claude/CLAUDE.md 1k
in the project directory: web-casoon-dev +4k · kestrel +8k · chartlet +400
? cache-rebuild look at it · notice · observed · v1
⇒ 202 rebuilds (60.1 M) are in sub-agent files, which wrote 100 % of their caches with
5 min while the main conversation wrote 100 % with 1 h; the log does not show
whether a setting reaches sub-agents; 147 of 349 unexpected rebuilds (42.4 M) follow
pauses over 1 h, which no cache lifetime covers
pause over 1 h █████████████████······· 42.4 M 147 rebuilds
5 min expired, main session ························ 0 0 rebuilds
5 min expired, sub-agents ████████████████████████ 60.1 M 202 rebuilds
cause unknown ························ 0 0 rebuilds
expected (model, compaction) ▌······················· 1.2 M 3 rebuilds
? long-session look at it · info · observed · v1
⇒ 22 sessions ran over a day, idle time included; only a change of topic inside one
makes a fresh session worth it, and the log does not show topics
runemark · 0f903e83 ████████████████████████ 155 h 500 requests
auditmysite · 1dadfa98 ██████████████████▏····· 117 h 116 requests
astro-site-files · 0f91e74f ██████████████████······ 116 h 389 requests
✓ big-context fine · notice · observed · v2
⇒ cheap while cached: 98 % of these tokens were cache reads; 39 % of these tokens sit
in 3 of 22 sessions; whether that length helped the task the log cannot show
kestrel · 21a6b55d ██████▌················· 360 M 509 requests
web-casoon · 2e02dbf0 ████▌··················· 250 M 355 requests
kestrel · 3ad3ed6b ████▏··················· 227 M 316 requests
19 other sessions ████████████████████████ 1.32 G 2,076 requests
✓ compact-thrash nothing to report: 0 of 15 automatic compaction pairs came with under
100k of context growth (v1)