New: 98,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Audit State Machine

by ben-manes in ben-manes/caffeine

Audit explicit state machines (drain status, node lifecycle, async-value lifecycle) for illegal or missed transitions

Apache-2.0Auto-check passed

Install Audit State Machine

skills CLI
$ npx skills add ben-manes/caffeine --skill audit-state-machine -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install ben-manes/caffeine audit-state-machine --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/ben-manes/caffeine.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.claude/skills/audit-state-machine .claude/skills/audit-state-machine && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
audit-state-machine
GitHub stars
18k
Token cost
~1.1k tokens
SKILL.md length
548 words
Files
1
Skills in repo
33
Repo updated
First seen
Licence
Apache-2.0

At a glance

Audit explicit state machines (drain status, node lifecycle, async-value lifecycle) for illegal or missed transitions

  • Works in 4 steps: Lost wakeup: can the machine settle in… → Double schedule: can two threads both… → Opaque vs CAS staleness: reads are… → …
  • SKILL.md covers Machine 1: Drain status…, Machine 2: Node lifecycle, Machine 3: Async-value lifecycle and Output
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md

What it does

Audit State Machine is an agent skill from ben-manes/caffeine. Audit explicit state machines (drain status, node lifecycle, async-value lifecycle) for illegal or missed transitions

Its SKILL.md is about 1.1k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

The repository describes itself as: A high performance caching library for Java. The licence is Apache-2.0.

Example prompts

  • “/audit-state-machine”

Workflow steps

4 steps, taken from the first numbered list in SKILL.md.

  1. Lost wakeup: can the machine settle in IDLE while work remains buffered? Trace
  2. Double schedule: can two threads both schedule maintenance for the same epoch,
  3. Opaque vs CAS staleness: reads are opaque, transitions are CAS/release. For every
  4. Pacer coupling: rescheduleCleanUpIfIncomplete gates on `REQUIRED &&

What it can do on your machine

Read from SKILL.md and the folder at commit e972fb0. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md.

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Audit State Machine loads about 1.1k tokens when it runs. Until then it costs about 34 tokens; SKILL.md has 548 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~34
When it runs · the whole SKILL.md, loaded when a task matches
~1.1k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from ben-manes/caffeine at commit e972fb0, republished under its Apache-2.0 licence (© ben-manes). 548 words, ~1,118 tokens.

Download SKILL.mdSave it as .claude/skills/audit-state-machine/SKILL.md (or your agent's skills folder).
name
audit-state-machine
description
Audit explicit state machines (drain status, node lifecycle, async-value lifecycle) for illegal or missed transitions
context
fork
agent
auditor
disable-model-invocation
true

Audit the cache's explicit state machines for illegal transitions, missed transitions (lost wakeups), and ABA across transitions. The snapshot audits trace fields and methods one at a time; this one builds the full transition table for each machine and asks whether every reachable interleaving keeps the machine legal. A missed transition wedges the cache (work buffered, never drained); an illegal transition resurrects a dead node or strands a future.

The drain/maintenance path was recently changed ("assist maintenance directly when the write buffer is full"), so Machine 1 is the priority.

Machine 1: Drain status (priority)

States: IDLE, REQUIRED, PROCESSING_TO_IDLE, PROCESSING_TO_REQUIRED. Transition sites: afterWrite, scheduleAfterWrite, scheduleDrainBuffers, maintenance, rescheduleCleanUpIfIncomplete, performCleanUp. Access via drainStatusOpaque/drainStatusAcquire, casDrainStatus, setDrainStatusOpaque/setDrainStatusRelease.

Build the table: for each (state, event) pair — a write arrives, a read arrives, maintenance starts/ends, the pacer fires, the executor rejects, the buffer-full inline-assist path runs — what is the next state and who drives it? Then attack:

  1. Lost wakeup: can the machine settle in IDLE while work remains buffered? Trace the maintenance-exit CAS (PROCESSING_TO_IDLE → IDLE) against a concurrent scheduleAfterWrite that observed PROCESSING_TO_IDLE and CAS'd it to PROCESSING_TO_REQUIRED. Which write loses, and does the fallback (setDrainStatusOpaque(REQUIRED)) re-arm it?
  2. Double schedule: can two threads both schedule maintenance for the same epoch, or the inline-assist path run concurrently with an executor-scheduled drain?
  3. Opaque vs CAS staleness: reads are opaque, transitions are CAS/release. For every decision that gates scheduling, can the opaque read be stale in a way that drops a reschedule? Verify the PROCESSING_TO_IDLE → PROCESSING_TO_REQUIRED CAS and the maintenance-exit re-check close the window on all paths.
  4. Pacer coupling: rescheduleCleanUpIfIncomplete gates on REQUIRED && !pacer.isScheduled(). Can REQUIRED coexist with no scheduled pacer and no in-flight maintenance — i.e. the cache wedged until the next user operation happens to drive it?
Show full SKILL.md (256 more words)Show less

Machine 2: Node lifecycle

States: alive → retired → dead. Strictly unidirectional. Sites: makeDead, the retire paths, generated isAlive/isRetired/isDead, eviction cancellation in evictEntry, and reuse of expired-but-alive nodes in remap/compute.

  1. Can any path move dead → retired, dead → alive, or retired → alive? evictEntry's resurrect flag declines eviction before retirement. remap/compute may reuse an expired-but-alive node under synchronized(node); neither permits reversing a retired or dead state.
  2. On every exception or early-return in the compute and eviction paths, does the node land in a legal terminal state — never stuck retired with no one left to finish makeDead?
  3. Is weight / region accounting applied exactly once per transition — not twice on a retried path, not zero on an exception path?

Machine 3: Async-value lifecycle

An async entry's value is an incomplete future → completes (value | null | exception). Sites: isComputingAsync, ASYNC_EXPIRY, refreshes(), the refresh bit in writeTime (& 1L).

  1. Can an entry be treated as both computing-async and expired/evicted in a way that strands the future or the ASYNC_EXPIRY timestamp? (Historical: timestamp stuck after executor rejection.)
  2. The refresh-in-progress bit in writeTime and the refreshes() map: can they disagree — bit set but map entry gone, or vice versa — so a refresh is double-started or never cleared?

Output

For each finding: the interleaving (thread-by-thread), the illegal or missed transition, the observable consequence (wedged cache, lost notification, stranded future, resurrected dead node), and a Verification. Verify each interleaving is JMM-legal, not merely sequentially consistent. If a transition cannot be resolved statically, ESCALATE with a Fray skeleton — the drain machine is a prime Fray target.

© ben-manes, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in .claude/skills/audit-state-machine of ben-manes/caffeine.

Open the folder on GitHubat commit e972fb0

Compare with similar skills

Audit State Machine next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Audit State Machine compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Audit State Machine this skillben-manes/caffeine18k—~1.1kAutomated safety check: PassApache-2.0
Explaining Machine Learning Modelsforyourhealth111-pixel/Vibe-Skills3.6k—~383Automated safety check: PassMIT
Training Machine Learning Modelsjeremylongshore/tons-of-skills-marketplace2.8k1 repos~1kAutomated safety check: PassMIT
Engineering Features For Machine Learningjeremylongshore/tons-of-skills-marketplace2.8k—~1.1kAutomated safety check: PassMIT
Explaining Machine Learning Modelsjeremylongshore/tons-of-skills-marketplace2.8k1 repos~1kAutomated safety check: PassMIT
Deploying Machine Learning Modelsjeremylongshore/tons-of-skills-marketplace2.8k—~830Automated safety check: PassMIT

Similar skills

  • Explaining Machine Learning Models

    foryourhealth111-pixel/Vibe-Skills

    Explain trained machine learning models through feature attribution, local explanations, and behavior summaries.

    3.6k GitHub stars~383 tokensUpdated 1 mo ago
    Data & AnalyticsAuto-check passed

  • Training Machine Learning Models

    jeremylongshore/tons-of-skills-marketplace

    Build train machine learning models with automated workflows.

    2.8k GitHub starsUsed in 1 repo~1k tokens
    Data & AnalyticsAuto-check passed

  • Engineering Features For Machine Learning

    jeremylongshore/tons-of-skills-marketplace

    Execute create, select, and transform features to improve machine learning model performance.

    2.8k GitHub stars~1.1k tokensUpdated yesterday
    Data & AnalyticsAuto-check passed

  • Explaining Machine Learning Models

    jeremylongshore/tons-of-skills-marketplace

    Build this skill enables AI assistant to provide interpretability and explainability for machine learning models.

    2.8k GitHub starsUsed in 1 repo~1k tokens
    Data & AnalyticsAuto-check passed

  • Deploying Machine Learning Models

    jeremylongshore/tons-of-skills-marketplace

    Deploy this skill enables AI assistant to deploy machine learning models to production environments.

    2.8k GitHub stars~830 tokensUpdated yesterday
    Data & AnalyticsAuto-check passed

  • Evaluating Machine Learning Models

    jeremylongshore/tons-of-skills-marketplace

    Build this skill allows AI assistant to evaluate machine learning models using a comprehensive suite of metrics.

    2.8k GitHub stars~914 tokensUpdated yesterday
    Data & AnalyticsAuto-check passed

More from ben-manes/caffeine

All 33 skills in this repo

  • Caffeine Cache Optimization Experiments

    ben-manes/caffeine

    Runs controlled JMH experiments on the Caffeine cache to find shared contention and hot-path waste, then reviews correctness and returns a reviewable patch.

    18k GitHub stars~2.6k tokensUpdated 3 days ago
    Auto-check: notes

  • Git History Bug Audit

    ben-manes/caffeine

    Audits a module by walking its git history commit by commit, tracking unresolved issues forward, and reporting the ones that survive to HEAD as findings.

    18k GitHub stars~3.3k tokensUpdated 3 days ago
    Auto-check passed

  • Adversarial Codebase Audit

    ben-manes/caffeine

    Runs a hostile review of the Caffeine Java caching library with parallel subagents that get no design docs, then challenges and consolidates their findings.

    18k GitHub stars~1.9k tokensUpdated 3 days ago
    Auto-check: notes

  • Caffeine Performance Audit

    ben-manes/caffeine

    Audits the Caffeine cache source for hot-path costs such as allocations, contention and memory layout, reporting only findings tied to specific lines.

    18k GitHub stars~559 tokensUpdated 3 days ago
    Auto-check passed

  • Audit Sibling Divergence

    ben-manes/caffeine

    Compares code paths that should behave the same, such as sync and async cache methods, and requires a concrete scenario where the two observably disagree.

    18k GitHub stars~4.3k tokensUpdated 3 days ago
    Auto-check: notes

  • Climber Step Minimization

    ben-manes/caffeine

    Prices each step of the window climber algorithm by disabling it in turn, to find steps that no longer earn their keep and branches that no longer fire.

    18k GitHub stars~3k tokensUpdated 3 days ago
    Auto-check: notes

Questions about Audit State Machine

What does Audit State Machine do?

Audit explicit state machines (drain status, node lifecycle, async-value lifecycle) for illegal or missed transitions. Audit State Machine is an agent skill from ben-manes/caffeine.

How do I install Audit State Machine in Claude Code?

Run `npx skills add ben-manes/caffeine --skill audit-state-machine -a claude-code`. Or copy the skill folder (.claude/skills/audit-state-machine in ben-manes/caffeine) into .claude/skills/audit-state-machine in your project. Claude Code loads it when a task matches its description.

How do I install Audit State Machine in Codex?

Run `npx skills add ben-manes/caffeine --skill audit-state-machine -a codex`. Or copy the skill folder (.claude/skills/audit-state-machine in ben-manes/caffeine) into .agents/skills/audit-state-machine in your project. Codex loads it when a task matches its description.

Can I use Audit State Machine in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add ben-manes/caffeine --skill audit-state-machine -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/audit-state-machine, .gemini/skills/audit-state-machine, .github/skills/audit-state-machine and .opencode/skills/audit-state-machine in your project.

What does Audit State Machine need to run?

SKILL.md names no scripts, command-line tools or credentials: Audit State Machine is instructions for the agent only.

Does Audit State Machine access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Audit State Machine safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Audit State Machine use?

Audit State Machine is published under the Apache-2.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Audit State Machine use?

About 1.1k tokens (SKILL.md is roughly 4.5k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Audit State Machine?

Skills that share tags, products or a category with Audit State Machine: Explaining Machine Learning Models (foryourhealth111-pixel/Vibe-Skills, 3.6k stars), Training Machine Learning Models (jeremylongshore/tons-of-skills-marketplace, 2.8k stars), Engineering Features For Machine Learning (jeremylongshore/tons-of-skills-marketplace, 2.8k stars) and Explaining Machine Learning Models (jeremylongshore/tons-of-skills-marketplace, 2.8k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Audit State Machine?

ben-manes (a GitHub user) maintains it in ben-manes/caffeine, which has 17,880 GitHub stars. The repository holds 33 skills in this directory. The repository was last updated on October 6, 2026.

Source: ben-manes/caffeine on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.