Tessl • The management layer for AI agent skills

Blog Docs Log inGet started

SkillPublic

cost-aware-llm-pipeline

LLM API 使用成本优化模式 —— 基于任务复杂度的模型路由、预算跟踪、重试逻辑和提示缓存。

Quality

73%

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

Fix and improve this skill with Tessl

tessl review fix ./docs/zh-CN/skills/cost-aware-llm-pipeline/SKILL.md

Quality

Content

68%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, code-first skill whose four techniques are each demonstrated with mostly executable Python. Its main weaknesses are the missing validation/feedback loop in the batch-oriented composition workflow and some redundancy between the activation and applicability sections plus a rot-prone pricing table.

Suggestions

Add an explicit validation/feedback step to the '组合' pipeline — e.g. verify parse_result output, and on BudgetExceededError show the recovery path (downgrade to a cheaper model or halt the batch) — to lift workflow clarity past the batch cap of 3.

Merge '何时激活' and '适用场景' into a single section, since both list when/where the skill applies and the duplication wastes tokens.

Move the '价格参考（2025-2026）' table into a reference file (or label it as time-sensitive with a re-verification note), and complete the process() example with real client setup and token accounting so the composition is copy-paste executable.

Dimension	Reasoning	Score
Conciseness	The body is code-dense and lean with informative comments, but the '何时激活' and '适用场景' sections largely duplicate each other, and the dated '价格参考（2025-2026）' table places time-sensitive data outside any deprecated/old-patterns section — minor trims remain, fitting the efficient-but-trimmable anchor.	4 / 5
Actionability	select_model, CostTracker, and the prompt-caching snippet are executable, but the composition function uses placeholders ('estimated_items', 'input_tokens=...', undefined 'Config'/'client'/'BudgetExceededError') and call_with_retry omits 'import time' — mostly executable with minor gaps rather than copy-paste ready.	4 / 5
Workflow Clarity	The '组合' section presents a clear numbered 4-step sequence with a budget checkpoint, but batch pipelines are an explicit use case and the workflow has no result validation, no budget-exceeded feedback loop, and no per-item enforcement — the batch-operation cap holds workflow clarity at 3.	3 / 5
Progressive Disclosure	The skill is a single well-sectioned file with no bundle files to reference and all content arguably core; however, at ~183 lines the full pricing table and complete code examples could partly live in reference files, leaving minor organization gaps rather than ideal placement.	4 / 5
	Total	15 / 20 Passed

Description

66%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.

The description clearly communicates what the skill does with four concrete named techniques, but it entirely omits when to activate it, which caps its usefulness as a trigger. Trigger-term coverage is good though missing common user phrasings for cost reduction.

Suggestions

Add a 'Use when...' clause, e.g. 'Use when building applications that call LLM APIs, running batch pipelines, or needing to keep API spending within a budget' — this would raise completeness from 3 toward 5.

Include natural trigger synonyms users would actually say, such as '降低 API 费用', 'reduce LLM costs', 'budget limits', or 'cheaper model routing', to broaden trigger-term coverage.

Dimension	Reasoning	Score
Specificity	The description names the domain ("LLM API 使用成本优化模式") and lists four concrete capabilities ("模型路由、预算跟踪、重试逻辑和提示缓存"), matching the 'several specific actions, minor gaps' anchor rather than the comprehensive 5-anchor.	4 / 5
Completeness	The 'what' is clear — cost-optimization patterns with the four named techniques — but there is no 'Use when...' clause or equivalent trigger guidance anywhere, capping completeness at 3.	3 / 5
Trigger Term Quality	Terms like "LLM API", "成本优化", "预算跟踪", and "提示缓存" are natural phrases users would say, but common variations such as '降低费用', '省钱', 'reduce spending', or 'cheaper model' are missing, so coverage is good rather than comprehensive.	4 / 5
Distinctiveness Conflict Risk	The combination of model routing, budget tracking, and prompt caching is a fairly distinct niche within LLM development; only minor overlap risk exists with general LLM-pipeline or prompt-engineering skills.	4 / 5
	Total	15 / 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
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	15 / 16 Passed

Repository: xu-xiang/everything-claude-code-zh
Path: docs/zh-CN/skills/cost-aware-llm-pipeline/SKILL.md
Revision: dfbf946

Reviewed: 1 month ago

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.

cost-aware-llm-pipeline

Table of Contents

Is this your skill?