Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/zebbern/claude-code-guide/cachingnpx skills add zebbern/claude-code-guide --skill cachinggit clone --depth 1 https://github.com/zebbern/claude-code-guideWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00032 | $0.01584 |
| Opus 5 | $0.00016 | $0.00792 |
| Sonnet 5 | $0.00006 | $0.00317 |
| Haiku 4.5 | $0.00003 | $0.00158 |
Grade A, and why
caching scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 163 lines — stays where its author put it; the contents beside it link to each section on GitHub.
WHEN_TO_USE
- When implementing a cache layer (in-memory, Redis, CDN) for an API or service.
- When choosing TTL values or invalidation strategies for cached data.
- When designing cache key schemas to avoid collisions or stale-data bugs.
- When reviewing code that reads from or writes to any cache.
- When debugging stale data, cache stampedes, or inconsistent responses.
- When configuring TanStack Query
staleTime/gcTimefor client-side caching.
INVALIDATION
- [P0-MUST] Define an invalidation strategy for every cache. Stale data is worse than no cache.
- [P0-MUST] Invalidate caches when the underlying data changes — do not rely solely on TTL expiry.
- [P1-SHOULD] Prefer event-driven invalidation (on write/update/delete) over time-based expiry alone.
- [P1-SHOULD] Use cache versioning (include a version key) when data schemas change.
TTL_GUIDELINES
- [P1-SHOULD] Set TTLs based on data volatility: static config (hours/days), user profiles (minutes), real-time data (seconds or no cache).
- [P1-SHOULD] Use stale-while-revalidate: serve stale data immediately while refreshing in the background.
- [P2-MAY] Use shorter TTLs in development and longer TTLs in production.
CACHE_KEYS
- [P0-MUST] Include all query parameters that affect the result in the cache key.
- [P1-SHOULD] Use a consistent key format:
<entity>:<id>:<variant>(e.g.,user:123:profile,products:list:page=2). - [P1-SHOULD] Namespace keys by service or module to prevent collisions.
- [P2-MAY] Hash long or complex keys to keep storage efficient.
CACHE_LAYERS
- [P1-SHOULD] Use the appropriate cache layer for the use case:
| Layer | Best For | TTL Range |
|---|---|---|
| In-memory (Map, LRU) | Hot data, single-instance apps | Seconds to minutes |
| Redis / Memcached | Shared cache across instances, sessions | Minutes to hours |
| CDN / Edge | Static assets, public API responses | Hours to days |
| HTTP cache headers | Browser caching, API responses | Varies by resource |
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- yesterday First seen · 163 lines · 32 tokens per session scan A 0d4edff8dcc7
caching is a skill published in the GitHub repository zebbern/claude-code-guide (4,596 stars, last pushed 3d ago), licensed MIT. It adds 32 tokens to every session and 1,584 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
cdb-scan
Map this codebase into project memory — a code graph of every symbol and how they connect, plus a written profile of stack, layout, conventions and workflows. Re-run any time to refresh both in place. Use when memory is newly installed on an existing project, or when the project has changed enough that the stored map…
mckinsey-consultant
McKinsey顾问式问题解决系统。从商业问题出发,通过假设驱动的结构化分析方法,生成McKinsey风格研究报告和PPT。融合Problem Solving方法论、MECE原则、Issue Tree拆解、Hypotheses形成、Dummy Page设计、智能数据收集和专业PPT生成能力。.
exam-coach
Quiz and coach the user for the Anthropic Claude certification exams using this repository's blueprints and official exam guides. Use when the user asks to practice, be quizzed, drill a domain, take a mock exam, or prepare for the Associate, Developer, or Architect certifications.
security-claude
Skill "security-claude" from rahozosman/security-claude, covering security architecture & threat modeling intelligence, how this skill is organized (progressive disclosure), 1. pick a mode, 2. core method (applies to every mode) and 3. doing a focused review.
bridger
Coordinate with another Claude Code session over the bridge — discover peers, ask them, answer their questions — INSTEAD of guessing or asking the user. Trigger this the moment the task depends on something another repo's session knows: a dependency/library that changed and this code consumes it, an API or schema…
run-tests
Run the pytest suite, report pass/fail counts and coverage, and identify untested code. Use when the user asks to run tests, check test coverage, or verify that changes didn't break anything.