agent-radar · configured vs activated

Activation Gap — what you configured but never used

Overlays the configured side (solid lines) with the activated side (dashed lines) on the same five axes. The area between them is the capability waste zone.

Activation Gap Radar

CLAUDE.md Skills MCP Automation Context Hygiene
tradestation-monarch 69→ 44~ (user-space) 40→ N/A
Configured (filesystem fingerprint)Activated (session JSONL)

Biggest Gaps

Top items by configured vs activated gap. Large gap = configured but unused; that's the cheapest capability to fix.

1
UNDERUSED skills tradestation-monarch
`tradestation-monarch`: Skills configured but rarely triggered. Consider rewriting SKILL.md descriptions with explicit trigger scenarios.
· configured 100 · activated 0 · gap +100 (100%)
▾
CONFIGURED
Skills present 70/70
Found 4 SKILL.md file(s)
Lint: frontmatter & token hygiene 30/30
frontmatter compliant 4/4
ACTIVATED
Skill tool invocations 0/100
Skill never triggered (description may be weak, or no skills installed)
2
UNDERUSED context_hygiene tradestation-monarch
`tradestation-monarch`: CLAUDE.md / settings are well-structured, but sessions rarely use @ references. Build a `@path` habit on hot files to keep context focused.
· configured 70 · activated 19 · gap +51 (72%)
▾
CONFIGURED
User/Project split 40/40
Both project- and user-space config present (good split)
Shared vs personal config 0/30
Shared vs personal config not separated
Modular references 30/30
2 modular reference(s)
ACTIVATED
Read efficiency (low repetition) 19/50
83/133 reads were repeats (62.4%)
@ reference rate 0/50
2 @-mention(s) in 828 user message(s) (rate=0.00/msg)
3
UNDERUSED mcp tradestation-monarch
`tradestation-monarch`: You configured MCP servers but the invocation ratio is low. Audit which servers are never used and prune or reconsider.
· configured 18 · activated 0 · gap +18 (100%)
▾
CONFIGURED
MCP server count 18/50
1 MCP server(s)
MCP category breadth 0/50
Could not classify categories
ACTIVATED
MCP tool invocations 0/100
MCP server never invoked

Per-Target Detail

tradestation-monarch

D:\project\tradestation-monarch
69
DimensionConfiguredActivatedGap
CLAUDE.md 84 100 -16
Skills 100 0 +100
MCP 18 0 +18
Automation 75 100 -25
Context Hygiene 70 19 +51
CLAUDE.md 84 / 100 -16
CONFIGURED
CLAUDE.md exists 50/50
Found CLAUDE.md, .claude\CLAUDE.md
@import references 10/10
2 @ reference(s)
Lint: file size 15/15
6912 chars (within limit)
Iteration evidence 9/25
3 git commit(s) on CLAUDE.md; 0 iteration-loop content signal(s) (lessons-learned / do-not-repeat / dated rules)
ACTIVATED
Guidance effectiveness (low correction rate) 100/100
0/828 user messages contain corrections (0.0%) — high rate ⇒ CLAUDE.md not guiding
Skills 100 / 0 +100
CONFIGURED
Skills present 70/70
Found 4 SKILL.md file(s)
Lint: frontmatter & token hygiene 30/30
frontmatter compliant 4/4
ACTIVATED
Skill tool invocations 0/100
Skill never triggered (description may be weak, or no skills installed)
MCP 18 / 0 +18
CONFIGURED
MCP server count 18/50
1 MCP server(s)
MCP category breadth 0/50
Could not classify categories
ACTIVATED
MCP tool invocations 0/100
MCP server never invoked
Automation 75 / 100 -25
CONFIGURED
Hooks 30/30
Hooks configured
Subagents 30/30
7 subagent(s)
Custom slash commands 0/25
No custom commands
Plugins 15/15
Plugins detected
ACTIVATED
Subagent dispatches 100/100
41 subagent dispatch(es) via Agent tool
Context Hygiene 70 / 19 +51
CONFIGURED
User/Project split 40/40
Both project- and user-space config present (good split)
Shared vs personal config 0/30
Shared vs personal config not separated
Modular references 30/30
2 modular reference(s)
ACTIVATED
Read efficiency (low repetition) 19/50
83/133 reads were repeats (62.4%)
@ reference rate 0/50
2 @-mention(s) in 828 user message(s) (rate=0.00/msg)

BLIND SPOTscan only measures what's configured on disk. Pair with `agent-radar session` to see what actually fires inside Claude Code — the gap is your improvement headroom.

~ (user-space)

C:\Users\millerlai
40
DimensionConfiguredActivatedGap
CLAUDE.md 73 N/A —
Skills 70 N/A —
MCP 0 N/A —
Automation 15 N/A —
Context Hygiene 40 N/A —
CLAUDE.md 73 / N/A —
CONFIGURED
CLAUDE.md exists 50/50
Found .claude\CLAUDE.md
@import references 0/10
@import not used to split files
Lint: file size 15/15
1279 chars (within limit)
Iteration evidence 8/25
0 git commit(s) on CLAUDE.md; 2 iteration-loop content signal(s) (lessons-learned / do-not-repeat / dated rules)
ACTIVATED
No activation signals (no sessions logged?)
Skills 70 / N/A —
CONFIGURED
Skills present 70/70
Found 146 SKILL.md file(s)
Lint: frontmatter & token hygiene 0/30
frontmatter compliant 146/146, 82 ASCII-art / decorative banner violation(s), 114.0 oversize file(s)
ACTIVATED
No activation signals (no sessions logged?)
MCP 0 / N/A —
CONFIGURED
MCP server count 0/50
No MCP server configured
MCP category breadth 0/50
Could not classify categories
ACTIVATED
No activation signals (no sessions logged?)
Automation 15 / N/A —
CONFIGURED
Hooks 0/30
Hooks not in use
Subagents 0/30
No subagents defined
Custom slash commands 0/25
No custom commands
Plugins 15/15
Plugins detected
ACTIVATED
No activation signals (no sessions logged?)
Context Hygiene 40 / N/A —
CONFIGURED
User/Project split 40/40
Both project- and user-space config present (good split)
Shared vs personal config 0/30
Shared vs personal config not separated
Modular references 0/30
Not split into modules
ACTIVATED
No activation signals (no sessions logged?)

BLIND SPOTscan only measures what's configured on disk. Pair with `agent-radar session` to see what actually fires inside Claude Code — the gap is your improvement headroom.