context-budget

Name: context-budget
Author: affaan-m

4.7kinstalls35kforks

How to install context-budget?

affaan-m/everything-claude-code/context-budget

$npx -y skills add affaan-m/everything-claude-code --skill context-budget

Installs into the current project.

Summary

Audits Claude Code context window consumption across agents, skills, MCP servers, and rules. Identifies bloat, redundant components, and produces prioritized token-savings recommendations.

Files · 1

View on GitHub

SKILL.md

1	# Context Budget
2
3	Analyze token overhead across every loaded component in a Claude Code session and surface actionable optimizations to reclaim context space.
4
5	## When to Use
6
7	- Session performance feels sluggish or output quality is degrading
8	- You've recently added many skills, agents, or MCP servers
9	- You want to know how much context headroom you actually have
10	- Planning to add more components and need to know if there's room
11	- Running `/context-budget` command (this skill backs it)
12
13	## How It Works
14
15	### Phase 1: Inventory
16
17	Scan all component directories and estimate token consumption:
18
19	Agents (`agents/*.md`)
20	- Count lines and tokens per file (words × 1.3)
21	- Extract `description` frontmatter length
22	- Flag: files >200 lines (heavy), description >30 words (bloated frontmatter)
23
24	Skills (`skills/*/SKILL.md`)
25	- Count tokens per SKILL.md
26	- Flag: files >400 lines
27	- Check for duplicate copies in `.agents/skills/` — skip identical copies to avoid double-counting
28
29	Rules (`rules/*/.md`)
30	- Count tokens per file
31	- Flag: files >100 lines
32	- Detect content overlap between rule files in the same language module
33
34	MCP Servers (`.mcp.json` or active MCP config)
35	- Count configured servers and total tool count
36	- Estimate schema overhead at ~500 tokens per tool
37	- Flag: servers with >20 tools, servers that wrap simple CLI commands (`gh`, `git`, `npm`, `supabase`, `vercel`)
38
39	CLAUDE.md (project + user-level)
40	- Count tokens per file in the CLAUDE.md chain
41	- Flag: combined total >300 lines
42
43	### Phase 2: Classify
44
45	Sort every component into a bucket:
46
47	\| Bucket \| Criteria \| Action \|
48	\|--------\|----------\|--------\|
49	\| Always needed \| Referenced in CLAUDE.md, backs an active command, or matches current project type \| Keep \|
50	\| Sometimes needed \| Domain-specific (e.g. language patterns), not referenced in CLAUDE.md \| Consider on-demand activation \|
51	\| Rarely needed \| No command reference, overlapping content, or no obvious project match \| Remove or lazy-load \|
52
53	### Phase 3: Detect Issues
54
55	Identify the following problem patterns:
56
57	- Bloated agent descriptions — description >30 words in frontmatter loads into every Task tool invocation
58	- Heavy agents — files >200 lines inflate Task tool context on every spawn
59	- Redundant components — skills that duplicate agent logic, rules that duplicate CLAUDE.md
60	- MCP over-subscription — >10 servers, or servers wrapping CLI tools available for free
61	- CLAUDE.md bloat — verbose explanations, outdated sections, instructions that should be rules
62
63	### Phase 4: Report
64
65	Produce the context budget report:
66
67	```
68	Context Budget Report
69	═══════════════════════════════════════
70
71	Total estimated overhead: ~XX,XXX tokens
72	Context model: Claude Sonnet (200K window)
73	Effective available context: ~XXX,XXX tokens (XX%)
74
75	Component Breakdown:
76	┌─────────────────┬────────┬───────────┐
77	│ Component │ Count │ Tokens │
78	├─────────────────┼────────┼───────────┤
79	│ Agents │ N │ ~X,XXX │
80	│ Skills │ N │ ~X,XXX │
81	│ Rules │ N │ ~X,XXX │
82	│ MCP tools │ N │ ~XX,XXX │
83	│ CLAUDE.md │ N │ ~X,XXX │
84	└─────────────────┴────────┴───────────┘
85
86	WARNING: Issues Found (N):
87	[ranked by token savings]
88
89	Top 3 Optimizations:
90	1. [action] → save ~X,XXX tokens
91	2. [action] → save ~X,XXX tokens
92	3. [action] → save ~X,XXX tokens
93
94	Potential savings: ~XX,XXX tokens (XX% of current overhead)
95	```
96
97	In verbose mode, additionally output per-file token counts, line-by-line breakdown of the heaviest files, specific redundant lines between overlapping components, and MCP tool list with per-tool schema size estimates.
98
99	## Examples
100
101	Basic audit
102	```
103	User: /context-budget
104	Skill: Scans setup → 16 agents (12,400 tokens), 28 skills (6,200), 87 MCP tools (43,500), 2 CLAUDE.md (1,200)
105	Flags: 3 heavy agents, 14 MCP servers (3 CLI-replaceable)
106	Top saving: remove 3 MCP servers → -27,500 tokens (47% overhead reduction)
107	```
108
109	Verbose mode
110	```
111	User: /context-budget --verbose
112	Skill: Full report + per-file breakdown showing planner.md (213 lines, 1,840 tokens),
113	MCP tool list with per-tool sizes, duplicated rule lines side by side
114	```
115
116	Pre-expansion check
117	```
118	User: I want to add 5 more MCP servers, do I have room?
119	Skill: Current overhead 33% → adding 5 servers (~50 tools) would add ~25,000 tokens → pushes to 45% overhead
120	Recommendation: remove 2 CLI-replaceable servers first to stay under 40%
121	```
122
123	## Best Practices
124
125	- Token estimation: use `words × 1.3` for prose, `chars / 4` for code-heavy files
126	- MCP is the biggest lever: each tool schema costs ~500 tokens; a 30-tool server costs more than all your skills combined
127	- Agent descriptions are loaded always: e

People who liked this also viewed

6 picks

Type

context-budget

4.7kinstalls35kforks

View on GitHub

How to install context-budget?

affaan-m/everything-claude-code/context-budget

$npx -y skills add affaan-m/everything-claude-code --skill context-budget

Installs into the current project.

Summary

Audits Claude Code context window consumption across agents, skills, MCP servers, and rules. Identifies bloat, redundant components, and produces prioritized token-savings recommendations.

Files · 1

View on GitHub

SKILL.md

1	# Context Budget
2
3	Analyze token overhead across every loaded component in a Claude Code session and surface actionable optimizations to reclaim context space.
4
5	## When to Use
6
7	- Session performance feels sluggish or output quality is degrading
8	- You've recently added many skills, agents, or MCP servers
9	- You want to know how much context headroom you actually have
10	- Planning to add more components and need to know if there's room
11	- Running `/context-budget` command (this skill backs it)
12
13	## How It Works
14
15	### Phase 1: Inventory
16
17	Scan all component directories and estimate token consumption:
18
19	Agents (`agents/*.md`)
20	- Count lines and tokens per file (words × 1.3)
21	- Extract `description` frontmatter length
22	- Flag: files >200 lines (heavy), description >30 words (bloated frontmatter)
23
24	Skills (`skills/*/SKILL.md`)
25	- Count tokens per SKILL.md
26	- Flag: files >400 lines
27	- Check for duplicate copies in `.agents/skills/` — skip identical copies to avoid double-counting
28
29	Rules (`rules/*/.md`)
30	- Count tokens per file
31	- Flag: files >100 lines
32	- Detect content overlap between rule files in the same language module
33
34	MCP Servers (`.mcp.json` or active MCP config)
35	- Count configured servers and total tool count
36	- Estimate schema overhead at ~500 tokens per tool
37	- Flag: servers with >20 tools, servers that wrap simple CLI commands (`gh`, `git`, `npm`, `supabase`, `vercel`)
38
39	CLAUDE.md (project + user-level)
40	- Count tokens per file in the CLAUDE.md chain
41	- Flag: combined total >300 lines
42
43	### Phase 2: Classify
44
45	Sort every component into a bucket:
46
47	\| Bucket \| Criteria \| Action \|
48	\|--------\|----------\|--------\|
49	\| Always needed \| Referenced in CLAUDE.md, backs an active command, or matches current project type \| Keep \|
50	\| Sometimes needed \| Domain-specific (e.g. language patterns), not referenced in CLAUDE.md \| Consider on-demand activation \|
51	\| Rarely needed \| No command reference, overlapping content, or no obvious project match \| Remove or lazy-load \|
52
53	### Phase 3: Detect Issues
54
55	Identify the following problem patterns:
56
57	- Bloated agent descriptions — description >30 words in frontmatter loads into every Task tool invocation
58	- Heavy agents — files >200 lines inflate Task tool context on every spawn
59	- Redundant components — skills that duplicate agent logic, rules that duplicate CLAUDE.md
60	- MCP over-subscription — >10 servers, or servers wrapping CLI tools available for free
61	- CLAUDE.md bloat — verbose explanations, outdated sections, instructions that should be rules
62
63	### Phase 4: Report
64
65	Produce the context budget report:
66
67	```
68	Context Budget Report
69	═══════════════════════════════════════
70
71	Total estimated overhead: ~XX,XXX tokens
72	Context model: Claude Sonnet (200K window)
73	Effective available context: ~XXX,XXX tokens (XX%)
74
75	Component Breakdown:
76	┌─────────────────┬────────┬───────────┐
77	│ Component │ Count │ Tokens │
78	├─────────────────┼────────┼───────────┤
79	│ Agents │ N │ ~X,XXX │
80	│ Skills │ N │ ~X,XXX │
81	│ Rules │ N │ ~X,XXX │
82	│ MCP tools │ N │ ~XX,XXX │
83	│ CLAUDE.md │ N │ ~X,XXX │
84	└─────────────────┴────────┴───────────┘
85
86	WARNING: Issues Found (N):
87	[ranked by token savings]
88
89	Top 3 Optimizations:
90	1. [action] → save ~X,XXX tokens
91	2. [action] → save ~X,XXX tokens
92	3. [action] → save ~X,XXX tokens
93
94	Potential savings: ~XX,XXX tokens (XX% of current overhead)
95	```
96
97	In verbose mode, additionally output per-file token counts, line-by-line breakdown of the heaviest files, specific redundant lines between overlapping components, and MCP tool list with per-tool schema size estimates.
98
99	## Examples
100
101	Basic audit
102	```
103	User: /context-budget
104	Skill: Scans setup → 16 agents (12,400 tokens), 28 skills (6,200), 87 MCP tools (43,500), 2 CLAUDE.md (1,200)
105	Flags: 3 heavy agents, 14 MCP servers (3 CLI-replaceable)
106	Top saving: remove 3 MCP servers → -27,500 tokens (47% overhead reduction)
107	```
108
109	Verbose mode
110	```
111	User: /context-budget --verbose
112	Skill: Full report + per-file breakdown showing planner.md (213 lines, 1,840 tokens),
113	MCP tool list with per-tool sizes, duplicated rule lines side by side
114	```
115
116	Pre-expansion check
117	```
118	User: I want to add 5 more MCP servers, do I have room?
119	Skill: Current overhead 33% → adding 5 servers (~50 tools) would add ~25,000 tokens → pushes to 45% overhead
120	Recommendation: remove 2 CLI-replaceable servers first to stay under 40%
121	```
122
123	## Best Practices
124
125	- Token estimation: use `words × 1.3` for prose, `chars / 4` for code-heavy files
126	- MCP is the biggest lever: each tool schema costs ~500 tokens; a 30-tool server costs more than all your skills combined
127	- Agent descriptions are loaded always: e

People who liked this also viewed

6 picks

Type