context-budget

Audits Claude Code context window consumption across agents, skills, MCP servers, and rules. Identifies bloat, redundant components, and produces prioritized token-savings recommendations. Use when the context window is filling up too fast and the agents, skills, MCP servers, or rules consuming it need to be identified.

Quality

81%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Quality

Content

75%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A well-structured, highly actionable audit workflow with concrete thresholds, an executable snippet, and a clear report format. The main weakness is token efficiency — the persisted-record section's repeated disclaimers and duplicated Best Practices entries add bulk without adding guidance value.

Suggestions

Move the persisted-record JSONL diagnostic and its caveat block into a reference file (e.g., references/persisted-bytes.md), keeping only a one-line pointer plus the snippet in the main body.

Deduplicate Best Practices against the phase content — 'MCP is the biggest lever (~500 tokens per tool)' and 'agent descriptions are loaded always' each restate facts already given in Phase 1/Phase 3.

Add a light validation checkpoint to the workflow (e.g., cross-check that per-component token sums match the report total, or re-scan after applying an optimization to confirm savings).

Dimension	Reasoning	Score
Conciseness	The phases, thresholds, and report template are tight, but the 'Persisted-record bytes' section carries ~50 lines of hedging caveats ('this does not find or reconnect a session', 'Neither category means "conversation"', 'These counts do not establish active context...') that pad the skill, and Best Practices restates facts already given inline (MCP ~500 tokens/tool and always-loaded agent descriptions each appear twice). Anchor 3: mostly efficient with some unnecessary explanation that could be tightened.	3 / 5
Actionability	Highly executable: numeric flag thresholds (>200 lines, >30-word descriptions, >20 tools/server, ~500 tokens per tool), a copy-paste-ready python3 heredoc snippet, a concrete classification table, and a filled-in report template with worked examples (basic, --verbose, pre-expansion). Not below 5 — examples cover the common cases with specific numbers.	5 / 5
Workflow Clarity	Four clearly sequenced phases (Inventory → Classify → Detect Issues → Report) with explicit per-component criteria. The destructive/batch cap does not apply since the audit is read-only. Not a 5 because there is no validation checkpoint (e.g., verifying estimates or re-checking after changes) — the sequence is present but checkpoints are implicit.	4 / 5
Progressive Disclosure	No bundle files exist (references/, scripts/, assets/ all absent), and the single-file body is well-sectioned with clear headers and a table. Good structure overall; the minor gap is that the optional ~50-line persisted-record diagnostic and its caveat block is inlined in the main body where a reference file would keep the core audit leaner.	4 / 5
	Total	16 / 20 Passed

Description

87%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A strong description: concrete actions, an explicit 'Use when...' clause with natural trigger language, and a distinct niche. Minor room to tighten trigger synonym coverage and mention the CLAUDE.md/rules audit surface.

Dimension	Reasoning	Score
Specificity	Names the target surface ('across agents, skills, MCP servers, and rules') and lists several concrete actions — 'Audits... consumption', 'Identifies bloat, redundant components', 'produces prioritized token-savings recommendations'. Not a 5 because coverage has minor gaps (e.g., CLAUDE.md/rules chain auditing is part of the skill but unmentioned, and 'identifies bloat' is somewhat generic).	4 / 5
Completeness	Explicitly answers both: 'what' (audits consumption, identifies bloat, produces prioritized recommendations) and 'when' via a literal 'Use when the context window is filling up too fast and the agents, skills, MCP servers, or rules consuming it need to be identified' clause with concrete trigger phrases.	5 / 5
Trigger Term Quality	Includes natural phrases users would say — 'context window is filling up too fast', 'agents, skills, MCP servers, or rules' — plus the concrete noun 'token-savings'. Not a 5 because common variations like 'context is full', 'headroom', 'token usage', or 'context budget' are only partially covered.	4 / 5
Distinctiveness Conflict Risk	Occupies a clear niche (context-window auditing of loaded components) with distinct triggers ('filling up too fast', component-type enumeration); unlikely to fire for unrelated skills. Third-person voice throughout.	5 / 5
	Total	18 / 20 Passed

Validation

93%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
metadata_version	'metadata.version' is missing	Warning

	Total	15 / 16 Passed

Repository: affaan-m/ECC
Path: skills/context-budget/SKILL.md
Commit: c70874f

Reviewed: 1 day ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.