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 skills add arpitbbhayani/ape-skills --skill ape-cut-fluffgit clone --depth 1 https://github.com/arpitbbhayani/ape-skillsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/arpitbbhayani/ape-skills/ape-cut-fluff)<a href="https://agentmods.dev/skills/arpitbbhayani/ape-skills/ape-cut-fluff"><img src="https://agentmods.dev/badge/skills/arpitbbhayani/ape-skills/ape-cut-fluff/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/arpitbbhayani/ape-skills/ape-cut-fluff"><img src="https://agentmods.dev/badge/skills/arpitbbhayani/ape-skills/ape-cut-fluff.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What 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 5.1 | $0.00114 | $0.04723 |
| Opus 5 | $0.00057 | $0.02361 |
| Sonnet 5 | $0.00023 | $0.00945 |
| Haiku 4.5 | $0.00011 | $0.00472 |
Grade A, and why
ape-cut-fluff scanned grade A with 1 finding 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 11d ago.
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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
2. **A URL**: a string starting with `http://` or `https://`. Use `WebFetch` to download the content, then strip navigation, sidebars, footers, and comments before working on the body. If the URL points to a PDF (researc How it starts
The opening of the file, as written. The whole thing — 170 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Cut Fluff Skill
A surgical, single-pass editor. Walks the input top to bottom and removes every sentence, phrase, or word that delays the point without adding meaning. Cuts only -- does not reinterpret, restructure, or rewrite arguments. The author's voice, facts, and structural intent stay intact; the fat comes off.
The goal is the bare minimum word count that still lets a competent reader understand the concept and grasp the essence. If a word can be dropped without changing meaning, drop it. If a sentence can be shrunk into a clause without losing a fact, shrink it. If an idea, example, or piece of context appears more than once, keep only the strongest instance and cut the rest. Aim for prose so dense that every remaining word is doing work and every remaining idea is appearing for the first time.
What Counts as Fluff
The skill cuts the following on sight. Any phrase, sentence, or paragraph that matches one of these patterns is fluff:
- Hedging language: "it is important to note that", "it should be mentioned that", "it is worth pointing out that", "needless to say", "as a matter of fact", "to be honest", "in my opinion", "I think", "perhaps", "arguably" -- when they soften a claim the author is willing to make outright
- Empty disclaimers: "of course there are exceptions", "this is a simplification", "your mileage may vary", "obviously this depends on context", "every system is different" -- if the disclaimer does not name a specific exception, cut it; if it names one, keep only the naming clause
- Throat-clearing intros: "In this post we will...", "Before we begin...", "Let us first understand...", "As we all know...", "To start with...", "First of all..." -- any windup that delays the first real sentence
- Cliché openers: "In today's fast-paced world...", "In the age of AI...", "As engineers, we all know...", "We live in a world where..." -- generic openers that could front any blog; cut and let the first concrete claim be the opener
- Empty subject openers: "There is...", "There are...", "It is...that...", "What is interesting is that...", "The thing about X is that..." -- strip the placeholder and let the real subject lead. Example: "There are many systems that fail under load" → "Many systems fail under load."
- Filler transitions: "Now that we understand X, let us look at Y", "With that said...", "Having said that...", "Moving on...", "That being said...", "On a related note..."
- Self-reference and meta-structure: "in this section", "in the next section we will", "this article will cover", "as we will see", "as discussed earlier", "as mentioned above", "more on this later", "we will return to this", "(spoiler alert)" -- meta-commentary about the structure of the post that adds no information
- Summarising conclusions: "As we can see...", "In summary...", "To wrap up...", "In conclusion...", "What this essentially means is..." -- especially when the next sentence repeats what was just said
- Rhetorical-question tails: "Right?", "Isn't it?", "Make sense?", "You with me?" -- performative, not informative
- Intro→body→conclusion symmetry: the "tell them what you will tell them, tell them, tell them what you told them" pattern; the previewing intro and the summarising closer are usually both fluff and the body carries the argument on its own; cut whichever the body makes redundant, keep the one that earns its place
- Padded constructions: "in order to" → "to", "due to the fact that" → "because", "at this point in time" → "now", "a large number of" → "many", "the majority of" → "most", "in the event that" → "if", "with regard to" → "about", "for the purpose of" → "to"
- Nominalisations: verbs buried inside noun phrases that bloat the sentence; restore the verb. "make a decision" → "decide", "perform an analysis" → "analyse", "carry out an evaluation" → "evaluate", "give consideration to" → "consider", "reach a conclusion" → "conclude", "have a discussion about" → "discuss", "is in agreement with" → "agrees with"
- Filler qualifiers: "basically", "essentially", "really", "actually", "quite", "very", "just", "simply", "literally", "definitely", "totally", "absolutely" -- when they add no information
- Adjective and intensifier stacks: "extremely fast", "incredibly difficult", "super important", "really critical", "absolutely essential", "highly significant" -- the noun or adjective carries the weight alone; cut the intensifier or replace the pair with a single stronger word
- Redundant pairs: "end result" → "result", "future plans" → "plans", "completely eliminate" → "eliminate", "advance planning" → "planning", "added bonus" → "bonus", "first introduced" → "introduced", "past history" → "history", "free gift" → "gift"
- Restatement and repetition: any idea, fact, definition, or framing that the post states more than once -- adjacent sentences that say the same thing, cross-paragraph repeats where the same point reappears sections later, definitional repetition where the same term or acronym is defined twice, and heading-body restatement where a heading and its first sentence carry the same claim ("Why caching matters" → "Caching matters because..."); keep the strongest mention and cut the rest, or collapse overlapping mentions into one sharper sentence
- Redundant examples and analogies: when the post offers two or three examples, analogies, or illustrations of the same point, keep the strongest one and cut the others; do not preserve a parallel example just because the author wrote it. The rule of thumb is: if the second example does not introduce a new failure mode, a new trade-off, or a new edge case the first did not surface, it is redundant
- Redundant background and context: setup, history, or motivation that the reader already has from earlier in the post, from the title, or from the heading; cut prerequisite explanations that the post itself just covered, recaps that exist only because the author is afraid the reader has forgotten, and "as a refresher" passages that re-explain something stated a few paragraphs above
- Redundant elaboration: a claim followed by a longer paragraph that says the same thing in more words without adding mechanism, evidence, or consequence; keep the crisp claim, cut the elaboration. If the elaboration introduces a number, a system name, a failure mode, or a trade-off the claim lacked, keep the new fact and cut the rest
- Idea redundancy across the whole post: the most expensive form of redundancy. Two paragraphs in different sections that make the same argument from slightly different angles; a sidebar that repeats the main thread; a "key takeaway" box that restates the section it sits inside. During the first read, flag every place the same idea reappears; in the cutting pass, keep one canonical mention -- usually the earliest or the most concrete -- and cut every echo, even if the echoes are well-written
- Dead sentences: any sentence that survives only because the sentence before it already carried the point
- AI-slop framing: overly balanced "On one hand... on the other hand..." paragraphs where one side is obviously the answer; symmetric conclusions that restate the intro; generic exhortations ("This is a powerful technique that engineers should consider when designing systems")
- Code-narration: prose right after a fenced code block that re-describes what the code does line by line; cut the narration and keep only sentences that add intent, trade-offs, or non-obvious behaviour the code itself does not show
- Ceremonial closers: "Thank you for reading", "Hope you found this useful", "Stay tuned for more" -- cut entirely
- Non-load-bearing filler words: individual words whose removal does not change meaning -- redundant articles ("the", "a", "an") where the sentence parses without them, auxiliary verbs ("is being used" → "uses", "can be seen to" → ""), expletive subjects, relative pronouns where the clause still works ("the system that we built" → "the system we built"), prepositional padding ("the field of distributed systems" → "distributed systems"), possessive "of" chains ("the design of the system" → "the system design"), and connective tissue ("and so", "and then", "but also") that pads without joining genuinely distinct ideas. Read each sentence and ask: which words can I delete and still leave the same claim standing? Delete them.
- Verbose phrasings: prefer the shorter equivalent on sight -- "is able to" → "can", "has the ability to" → "can", "in spite of the fact that" → "though", "despite the fact that" → "though", "the reason is because" → "because", "during the course of" → "during", "on a daily basis" → "daily", "a number of" → "some/many", "a variety of" → "various", "as a means of" → "to", "the question as to whether" → "whether"
- Compressible sentence frames: long-form sentence frames that compress into a phrase -- "It is important to understand that X" → "X", "What we need to do is X" → "X" or "Do X", "The way that this works is X" → "It works by X", "One thing to keep in mind is X" → "X". The leading frame is fluff; the embedded clause is the point
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.
- 11d ago First seen · 170 lines · 114 tokens per session scan A 66c1dd9ffe43
ape-cut-fluff is a skill published in the GitHub repository arpitbbhayani/ape-skills (40 stars, last pushed 2d ago), licensed MIT. It adds 114 tokens to every session and 4,723 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
offensive-c2-frameworks
Command and Control framework deployment, configuration, and operational tradecraft for red team engagements. Covers Cobalt Strike (malleable C2 profiles, Beacon types HTTP/HTTPS/DNS/SMB, Beacon Object Files for in-memory execution, sleep and jitter tuning, named pipe pivoting), Sliver (implant generation across…
offensive-crypto-attacks
Systematic methodology for identifying and exploiting cryptographic implementation weaknesses in real-world applications. Covers padding oracle attacks against CBC-mode ciphers with PKCS7 padding (Vaudenay's original attack through modern padbuster automation), ECB mode exploitation including block cut-and-paste and…
offensive-parameter-pollution
HTTP parameter pollution (HPP) checklist: duplicate parameter injection, backend vs frontend parsing differences, WAF bypass via HPP, server-side vs client-side HPP, and practical exploitation patterns. Use when testing web applications for parameter handling flaws.
offensive-wifi
Wireless / 802.11 attack methodology for red team engagements and wireless security assessments. Covers monitor-mode setup, WPA/WPA2-PSK handshake capture and PMKID attacks, WPA3 SAE downgrade and Dragonblood, WPA-Enterprise (EAP) attacks (MSCHAPv2 cracking, EAP-TLS cert theft, evil-twin RADIUS), Karma / Known Beacons…
offensive-z-wave
Z-Wave attack methodology — sniffing with Z-Force / EZ-Wave / RTL-SDR + ZniffMobile, S0 (legacy) network-key derivation flaw and key reuse, S2 (modern) ECDH commissioning analysis, replay/injection on unauthenticated nodes, default-key brute-force on test deployments, and home-automation hub pivots. Use when targeting…
seo
Optimize for search engine visibility and ranking. Use when asked to "improve SEO", "optimize for search", "fix meta tags", "add structured data", "sitemap optimization", or "search engine optimization".