bgauryy/octocode

#!/usr/bin/env python3

Executable file

#!/usr/bin/env bash

#!/usr/bin/env bash

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

Get keys: [Tavily](https://app.tavily.com/) · [Serper](https://serper.dev/) · [Exa](https://dashboard.exa.ai/) · any one is enough, more gives redundancy/fallback.

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

- *Divergent vs. convergent thinking as distinct constructs* — J.P. Guilford's 1950 APA presidential address ("Creativity") launched modern creativity psychology and the divergent/convergent distinction.   Guilford established the construct

Guilford established the constructs, not a "never mix the phases" rule. [Stanford Encyclopedia of Philosophy — Creativity](https://plato.stanford.edu/entries/creativity/) · [Britannica — Divergent thinking](https://www.britannica.com/scienc

Keep the two attributions distinct when citing this method. - *Defer judgment / quantity-first while diverging* — Alex Osborn's brainstorming rules, *Applied Imagination* (1953). Attribution and date are the standard bibliographic reference

- *Defer judgment / quantity-first while diverging* — Alex Osborn's brainstorming rules, *Applied Imagination* (1953). Attribution and date are the standard bibliographic reference via secondary sources.   A primary/publisher-adjacent page 

A primary/publisher-adjacent page was not independently reachable as of 2026-07-11 (treat 1953 as standard, not independently re-confirmed here). [Brainstorming — Wikipedia](https://en.wikipedia.org/wiki/Brainstorming)

- *Combine/shift lens (SCAMPER)* — Bob Eberle, *SCAMPER: Games for Imagination Development* (1971), systematizing Osborn's idea-spurring checklist. [SCAMPER — Wikipedia](https://en.wikipedia.org/wiki/SCAMPER)

- Du et al., *Improving Factuality and Reasoning in Language Models through Multiagent Debate* — multiple LLM instances critique each other over rounds and converge on one shared answer; optimizes agreement/factuality. [arXiv:2305.14325](ht

- Wang et al., *Unleashing the Emergent Cognitive Synergy in Large Language Models: A Task-Solving Agent through Multi-Persona Self-Collaboration* (SPP) — one model simulates multiple input-dependent personas across turns, including creativ

- Tavily `/search`: `search_depth` (`basic`/`advanced`/`fast`/`ultra-fast`), `topic` (general/news/finance), `time_range`, `max_results` (0–20), `include_domains`/`exclude_domains`, `auto_parameters`, `start_date`/`end_date`. [Tavily API re

- Serper `/search`: `q`, `gl`, `hl`, `num`, `page`, `tbs` (recency via `qdr:d|w|m|y`), `autocorrect`, `location`. [serper.dev](https://serper.dev/) · [Serper params (LiteLLM)](https://docs.litellm.ai/docs/search/serper)

[Credential-related source omitted]

1. If `octocode-research` is installed, load it and run the needed Map / Validate / Investigate flow. 2. If it is not installed, point the user or agent to https://github.com/bgauryy/octocode/tree/main/skills/octocode-research.

2. If it is not installed, point the user or agent to https://github.com/bgauryy/octocode/tree/main/skills/octocode-research.

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

#!/usr/bin/env node

#!/usr/bin/env node

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

#!/usr/bin/env node

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

#!/usr/bin/env node

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

Live Chrome/CDP evidence for agents: network, console, perf, DOM/actionability, storage, HAR, screenshots, auth-gated pages. Static crawl → [`octocode-scraping`](https://github.com/bgauryy/octocode/tree/main/skills/octocode-scraping).

## Source Of Truth Use official DevTools Protocol pages for unfamiliar, optional, experimental, deprecated, or failing methods: `https://chromedevtools.github.io/devtools-protocol/tot/<Domain>/`. If Chrome is running, `/json/protocol` gives

Use official DevTools Protocol pages for unfamiliar, optional, experimental, deprecated, or failing methods: `https://chromedevtools.github.io/devtools-protocol/tot/<Domain>/`. If Chrome is running, `/json/protocol` gives exact local suppor

[Credential-related source omitted]

node <skill-dir>/scripts/cookie-bridge.mjs --i-understand-secrets \   --from-port 9333 --to-port 9222 --urls "https://app.example.com"

--from-port 9333 --to-port 9222 --urls "https://app.example.com"

# from existing CDP source → isolated target node <skill-dir>/scripts/cookie-bridge.mjs --i-understand-secrets \   --from-port 9333 --to-port 9222 --urls "https://app.example.com"

node <skill-dir>/scripts/cookie-bridge.mjs --i-understand-secrets    --from-port 9333 --to-port 9222 --urls "https://app.example.com"

--from-port 9333 --to-port 9222 --urls "https://app.example.com"

# profile launch (Chrome must not lock the profile) node <skill-dir>/scripts/cookie-bridge.mjs --i-understand-secrets \   --from-profile Default --to-port 9222 --urls "https://app.example.com"

node <skill-dir>/scripts/cookie-bridge.mjs --i-understand-secrets    --from-profile Default --to-port 9222 --urls "https://app.example.com"

--from-profile Default --to-port 9222 --urls "https://app.example.com"

[Credential-related source omitted]

[Credential-related source omitted]

| Events not firing, or `--new-tab <url>` misses network/script events | Tab loaded before listeners attached — use `--new-tab about:blank`, attach listeners, then call `Page.navigate` inside `run()` | | JavaScript dialog blocking all comma

| JavaScript dialog blocking all commands | Add dialog guard before navigate: `cdp.on('Page.javascriptDialogOpening', () => cdp.send('Page.handleJavaScriptDialog', { accept: true }))` — see Dialog guard in `cdp-agent.md` section 0 | | URL w

| URL with `?` or `&` fails in zsh | Always quote the URL: `--url "http://..."` |

| network/console observation, performance, web vitals, DOM/a11y, heap, security | `references/script-patterns-observe.md` | | storage, consent, full-audit composition | `references/script-patterns-special.md` | | HAR export, API/curl repla

| HAR export, API/curl replay, token budgets | `references/har-capture.md` |

| **Opt-out** | `CDP_NO_STEALTH=1` or `cdp-runner.mjs --no-stealth` (debug only) | | **Debug only** | `CDP_STEALTH_ALLOW_FAIL=1` to log failures without exit; `CDP_SKIP_STEALTH_VERIFY=1` to skip verify | | **Navigation** | `--new-tab https:

| **Debug only** | `CDP_STEALTH_ALLOW_FAIL=1` to log failures without exit; `CDP_SKIP_STEALTH_VERIFY=1` to skip verify | | **Navigation** | `--new-tab https://…` opens `about:blank` first, patches, verifies, then navigates | | **Attached ta

| **Navigation** | `--new-tab https://…` opens `about:blank` first, patches, verifies, then navigates |

Executable file

Executable file

//   node skills/octocode-chrome-devtools/scripts/cdp-sandbox.mjs \ //     skills/octocode-chrome-devtools/scripts/cdp-checks/affiliates-stealth-probe.mjs \ //     --port 9222 --new-tab "https://affiliates.walmart.com/" --timeout 60000 --ke

//     skills/octocode-chrome-devtools/scripts/cdp-checks/affiliates-stealth-probe.mjs \ //     --port 9222 --new-tab "https://affiliates.walmart.com/" --timeout 60000 --keep-tab //

//     --port 9222 --new-tab "https://affiliates.walmart.com/" --timeout 60000 --keep-tab

const TARGET = process.env.AFFILIATES_PROBE_URL ?? 'https://affiliates.walmart.com/';

#!/usr/bin/env node

const response = await fetch(url, init);

Executable file

Executable file

#!/usr/bin/env node

#!/usr/bin/env node

Executable file

#!/usr/bin/env node

Executable file

const API_URL = 'https://example.test/api/data';

const html = `<script>fetch('${API_URL}').then(r=>r.json()).then(j=>document.body.textContent=JSON.stringify(j)).catch(e=>document.body.textContent=e.message)</script>`;

Executable file

const SLOW_MS = Number(process.env.NET_SLOW_MS || 1000); const API_URL = 'https://example.test/api/measure';

const API_URL = 'https://example.test/api/measure';

const API_BODY = JSON.stringify({ ok: true, id: 'net-measure' }); const FIXTURE = `data:text/html,${encodeURIComponent(`<!doctype html><script> fetch('${API_URL}').then(r=>r.json()).then(j=>{document.title='ok';document.body.textContent=JSO

const FIXTURE = `data:text/html,${encodeURIComponent(`<!doctype html><script> fetch('${API_URL}').then(r=>r.json()).then(j=>{document.title='ok';document.body.textContent=JSON.stringify(j)}).catch(e=>{document.body.textContent=e.message}); 

fetch('${API_URL}').then(r=>r.json()).then(j=>{document.title='ok';document.body.textContent=JSON.stringify(j)}).catch(e=>{document.body.textContent=e.message});

fetch('https://example.test/missing').catch(()=>{});

Executable file

Executable file

Executable file

const TARGET_URL = process.env.STEALTH_CHECK_URL ?? 'https://bot.sannysoft.com/';

#!/usr/bin/env node

Executable file

const FIXTURE_ORIGIN = 'https://storage.test/measure';

};   const pageHost = (() => {     try { return new URL(cdp.targetInfo?.url || storage.url || 'http://local').hostname; }

const pageHost = (() => {     try { return new URL(cdp.targetInfo?.url || storage.url || 'http://local').hostname; }     catch { return ''; }

try { return new URL(cdp.targetInfo?.url || storage.url || 'http://local').hostname; }

#!/usr/bin/env node

Executable file

#!/usr/bin/env node

async function cdpHttp(path, method = 'GET') {   const res = await fetch(`http://localhost:${PORT}${path}`, { method, signal: AbortSignal.timeout(5000) });

const res = await fetch(`http://localhost:${PORT}${path}`, { method, signal: AbortSignal.timeout(5000) });

async function closeTab(id)       {   try {     const res = await fetch(`http://localhost:${PORT}/json/close/${id}`, { signal: AbortSignal.timeout(3000) });

try {     const res = await fetch(`http://localhost:${PORT}/json/close/${id}`, { signal: AbortSignal.timeout(3000) });     return res.ok;

const res = await fetch(`http://localhost:${PORT}/json/close/${id}`, { signal: AbortSignal.timeout(3000) });

#!/usr/bin/env node

Executable file

const requests = new Map();   cdp.on('Network.requestWillBeSent', ({ requestId, request }) => {     requests.set(requestId, { url: request.url, method: request.method, start: Date.now() });

cdp.on('Network.requestWillBeSent', ({ requestId, request }) => {     requests.set(requestId, { url: request.url, method: request.method, start: Date.now() });   });

requests.set(requestId, { url: request.url, method: request.method, start: Date.now() });

});   cdp.on('Network.responseReceived', ({ requestId, response }) => {     const r = requests.get(requestId);

cdp.on('Network.responseReceived', ({ requestId, response }) => {     const r = requests.get(requestId);     if (!r) return;

const r = requests.get(requestId);

});   cdp.on('Network.loadingFailed', ({ requestId, errorText, blockedReason }) => {     const r = requests.get(requestId);

cdp.on('Network.loadingFailed', ({ requestId, errorText, blockedReason }) => {     const r = requests.get(requestId);     console.log(`[NETWORK_FAILED] ${r?.url ?? 'unknown'}: ${blockedReason ? `blocked:${blockedReason}` : errorText}`);

const r = requests.get(requestId);

#!/usr/bin/env node

*  * Optional:  *   --urls "https://a.com,https://b.com"   filter / setCookies URL hints

* Optional:  *   --urls "https://a.com,https://b.com"   filter / setCookies URL hints  *   --export-storage-state <path>          write jar (0600) without logging values

*   --urls "https://a.com,https://b.com"   filter / setCookies URL hints

console.error(`Usage: node cookie-bridge.mjs --i-understand-secrets --to-port 9222 \\   (--from-port 9333 | --from-profile Default | --from-storage-state state.json) \\   [--urls "https://example.com"] [--export-storage-state out.json] [--d

[--urls "https://example.com"] [--export-storage-state out.json] [--dry-run]

async function cdpHttp(port, path, method = 'GET') {   const res = await fetch(`http://127.0.0.1:${port}${path}`, {

const res = await fetch(`http://127.0.0.1:${port}${path}`, {

#!/usr/bin/env node

Executable file

#!/usr/bin/env node

Executable file

Executable file

#!/usr/bin/env node

async function checkRunning() {   try {     const res = await fetch(`http://localhost:${PORT}/json/version`, { signal: AbortSignal.timeout(2000) });

try {     const res = await fetch(`http://localhost:${PORT}/json/version`, { signal: AbortSignal.timeout(2000) });     if (res.ok) return await res.json();

const res = await fetch(`http://localhost:${PORT}/json/version`, { signal: AbortSignal.timeout(2000) });

const chromePath = CHROME_PATH || findChrome(); if (!chromePath) err('Chrome not found. Install Google Chrome from https://www.google.com/chrome/ or pass --chromePath <path>');

if (!chromePath) err('Chrome not found. Install Google Chrome from https://www.google.com/chrome/ or pass --chromePath <path>');

[Credential-related source omitted]

#!/usr/bin/env node

results.push({ session, url, status: res.status, ok: parsed?.ok ?? false, sessionDir: parsed?.sessionDir ?? null, resultPath }); } fetchSession('https://chromedevtools.github.io/devtools-protocol/', 'cdp-root');

} fetchSession('https://chromedevtools.github.io/devtools-protocol/', 'cdp-root'); for (const domain of domains) fetchSession(`https://chromedevtools.github.io/devtools-protocol/tot/${domain}`, `cdp-${domain}`);

fetchSession('https://chromedevtools.github.io/devtools-protocol/', 'cdp-root');

for (const domain of domains) fetchSession(`https://chromedevtools.github.io/devtools-protocol/tot/${domain}`, `cdp-${domain}`);

#!/usr/bin/env node

Executable file

Executable file

| What breaks the build? | `node scripts/style-lint.mjs docs/` — exit 1 on ERROR | | Is the word list current? | `node scripts/refresh-word-list.mjs --dry-run` | | Has the guide changed since? | [What's new](https://developers.google.com/st

| Has the guide changed since? | [What's new](https://developers.google.com/style/whats-new), then `references/style-sources.md` |

| Resource | Use for | |---|---| | [Google developer documentation style guide](https://developers.google.com/style) | The upstream guide these references restate |

|---|---| | [Google developer documentation style guide](https://developers.google.com/style) | The upstream guide these references restate | | [Highlights](https://developers.google.com/style/highlights) | The short list of rules that cove

| [Google developer documentation style guide](https://developers.google.com/style) | The upstream guide these references restate |

| [Highlights](https://developers.google.com/style/highlights) | The short list of rules that cover most reviews |

| [What's new](https://developers.google.com/style/whats-new) | The guide's changelog — the drift check for every reference here |

| [Word list](https://developers.google.com/style/word-list) | The upstream source of `assets/google-word-list.tsv` |

| [Diátaxis](https://diataxis.fr/) | The doc-type framework behind `references/diataxis.md` |

| [agents.md](https://agents.md/) | The spec behind `references/agents-md.md` |

This repository vendors the skill at `skills/octocode-documentation`. Google publishes its style guide content under [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/); this skill restates the rules and keeps the word-list guidance s

slug	title	owner	url highlights	Highlights	style-index.md	https://developers.google.com/style/highlights

highlights	Highlights	style-index.md	https://developers.google.com/style/highlights

philosophy	Philosophy of this guide	style-index.md	https://developers.google.com/style/philosophy

tone	Voice and tone	style-voice.md	https://developers.google.com/style/tone

person	Second person	style-voice.md	https://developers.google.com/style/person

voice	Active voice	style-voice.md	https://developers.google.com/style/voice

tense	Present tense	style-voice.md	https://developers.google.com/style/tense

anthropomorphism	Anthropomorphism	style-voice.md	https://developers.google.com/style/anthropomorphism

prescriptive-documentation	Prescriptive documentation	style-voice.md	https://developers.google.com/style/prescriptive-documentation

sentence-structure	Sentence structure	style-voice.md	https://developers.google.com/style/sentence-structure

paragraph-structure	Paragraphs	style-voice.md	https://developers.google.com/style/paragraph-structure

articles	Articles (a, an, the)	style-grammar.md	https://developers.google.com/style/articles

pronouns	Pronouns	style-grammar.md	https://developers.google.com/style/pronouns

possessives	Possessives	style-grammar.md	https://developers.google.com/style/possessives

pluralization	Pluralization	style-grammar.md	https://developers.google.com/style/pluralization

prepositions	Prepositions	style-grammar.md	https://developers.google.com/style/prepositions

contractions	Contractions	style-grammar.md	https://developers.google.com/style/contractions

word-list	Word list	style-words.md	https://developers.google.com/style/word-list

spelling	Spelling (redirects to the word list)	style-words.md	https://developers.google.com/style/spelling

abbreviations	Abbreviations	style-abbreviations.md	https://developers.google.com/style/abbreviations

jargon	Jargon	style-abbreviations.md	https://developers.google.com/style/jargon

translation	Global audience	style-global.md	https://developers.google.com/style/translation

accessibility	Accessibility	style-global.md	https://developers.google.com/style/accessibility

inclusive-documentation	Inclusive language	style-inclusive.md	https://developers.google.com/style/inclusive-documentation

headings	Headings and titles	style-structure.md	https://developers.google.com/style/headings

lists	Lists	style-structure.md	https://developers.google.com/style/lists

procedures	Procedures	style-procedures.md	https://developers.google.com/style/procedures

notices	Notes and other notices	style-blocks.md	https://developers.google.com/style/notices

tables	Tables	style-blocks.md	https://developers.google.com/style/tables

footnotes	Footnotes	style-blocks.md	https://developers.google.com/style/footnotes

images	Figures and other images	style-images.md	https://developers.google.com/style/images

text-formatting	Text-formatting summary	style-format.md	https://developers.google.com/style/text-formatting

capitalization	Capitalization	style-format.md	https://developers.google.com/style/capitalization

italics-terms	Italics with terms	style-format.md	https://developers.google.com/style/italics-terms

markdown	Markdown versus HTML	style-format.md	https://developers.google.com/style/markdown

semantic-tagging	HTML and semantic tagging	style-format.md	https://developers.google.com/style/semantic-tagging

html-formatting	HTML formatting	style-format.md	https://developers.google.com/style/html-formatting

filenames	Filenames	style-format.md	https://developers.google.com/style/filenames

colons	Colons	style-punctuation.md	https://developers.google.com/style/colons

commas	Commas	style-punctuation.md	https://developers.google.com/style/commas

dashes	Dashes	style-punctuation.md	https://developers.google.com/style/dashes

ellipses	Ellipses	style-punctuation.md	https://developers.google.com/style/ellipses

hyphens	Hyphens	style-punctuation.md	https://developers.google.com/style/hyphens

parentheses	Parentheses	style-punctuation.md	https://developers.google.com/style/parentheses

periods	Periods and end punctuation	style-punctuation.md	https://developers.google.com/style/periods

quotation-marks	Quotation marks	style-punctuation.md	https://developers.google.com/style/quotation-marks

semicolons	Semicolons	style-punctuation.md	https://developers.google.com/style/semicolons

slashes	Slashes	style-punctuation.md	https://developers.google.com/style/slashes

format-examples	Examples (how to introduce one)	style-punctuation.md	https://developers.google.com/style/format-examples

numbers	Numbers	style-numbers.md	https://developers.google.com/style/numbers

dates-times	Dates and times	style-numbers.md	https://developers.google.com/style/dates-times

units-of-measure	Units of measurement	style-numbers.md	https://developers.google.com/style/units-of-measure

phone-numbers	Phone numbers	style-numbers.md	https://developers.google.com/style/phone-numbers

mathematical-notation	Mathematical notation	style-numbers.md	https://developers.google.com/style/mathematical-notation

code-in-text	Code in text	style-code.md	https://developers.google.com/style/code-in-text

code-samples	Code samples	style-code.md	https://developers.google.com/style/code-samples

code-syntax	Command-line syntax	style-cli.md	https://developers.google.com/style/code-syntax

placeholders	Placeholder formatting	style-cli.md	https://developers.google.com/style/placeholders

examples	Example domains and names	style-examples.md	https://developers.google.com/style/examples

ui-elements	UI elements and interaction	style-ui.md	https://developers.google.com/style/ui-elements

cross-references	Cross-references and linking	style-links.md	https://developers.google.com/style/cross-references

headings-targets	Headings as link targets	style-links.md	https://developers.google.com/style/headings-targets

timeless-documentation	Timeless documentation	style-claims.md	https://developers.google.com/style/timeless-documentation

future	Future features	style-claims.md	https://developers.google.com/style/future

excessive-claims	Excessive claims	style-claims.md	https://developers.google.com/style/excessive-claims

other-sources	Third-party content	style-claims.md	https://developers.google.com/style/other-sources

trademarks	Trademarks	style-claims.md	https://developers.google.com/style/trademarks

product-names	Product names	style-claims.md	https://developers.google.com/style/product-names

api-reference-comments	API reference code comments	style-api.md	https://developers.google.com/style/api-reference-comments

reference-verbs	Verbs in reference documents	style-api.md	https://developers.google.com/style/reference-verbs

# Google developer documentation style guide word list — https://developers.google.com/style/word-list

currently	avoid	Avoid because this word is implied. The word can also prematurely disclose product or feature strategy or inappropriately imply that a product or feature might change. See also as of this writing and presently. Recommended: 

custom mode VPC network	usage	Not custom mode network. curl	usage	Not cURL. For information about when to use code format, see Items that are sometimes in code font. Cyber Monday	avoid	Avoid unless explicitly referring to an event in the US

curl	usage	Not cURL. For information about when to use code format, see Items that are sometimes in code font.

Load when creating or updating `AGENTS.md`, nested agent instructions, or a `CLAUDE.md` entrypoint. Spec: [agents.md](https://agents.md/).

Load when writing or reviewing human-facing docs. Framework: [Diátaxis](https://diataxis.fr/). One type per page; cross-link the others.

Upstream: [Abbreviations](https://developers.google.com/style/abbreviations) · [Jargon](https://developers.google.com/style/jargon). Verify a disputed or missing rule against the live page → `references/style-sources.md`.

Upstream: [API reference code comments](https://developers.google.com/style/api-reference-comments) · [Verbs in reference documents](https://developers.google.com/style/reference-verbs). Verify a disputed or missing rule against the live pa

Upstream: [Notes and other notices](https://developers.google.com/style/notices) · [Tables](https://developers.google.com/style/tables) · [Footnotes](https://developers.google.com/style/footnotes). Verify a disputed or missing rule against 

Upstream: [Timeless documentation](https://developers.google.com/style/timeless-documentation) · [Future features](https://developers.google.com/style/future) · [Excessive claims](https://developers.google.com/style/excessive-claims) · [Thi

Upstream: [Command-line syntax](https://developers.google.com/style/code-syntax) · [Placeholder formatting](https://developers.google.com/style/placeholders). Verify a disputed or missing rule against the live page → `references/style-sourc

| Case | Code font when | Plain when | |---|---|---| | CLI utility | it's the command itself: `gcloud`, `curl` | it's the project or product: "the curl project website" |

|---|---|---| | CLI utility | it's the command itself: `gcloud`, `curl` | it's the project or product: "the curl project website" | | Boolean | you mean the literal value `true` | you describe the evaluation: "the condition is false" |

| CLI utility | it's the command itself: `gcloud`, `curl` | it's the project or product: "the curl project website" |

Upstream: [Code in text](https://developers.google.com/style/code-in-text) · [Code samples](https://developers.google.com/style/code-samples). Verify a disputed or missing rule against the live page → `references/style-sources.md`.

Upstream: [Example domains and names](https://developers.google.com/style/examples). Verify a disputed or missing rule against the live page → `references/style-sources.md`.

Upstream: [Text-formatting summary](https://developers.google.com/style/text-formatting) · [Capitalization](https://developers.google.com/style/capitalization) · [Italics with terms](https://developers.google.com/style/italics-terms) · [Mar

Upstream: [Global audience](https://developers.google.com/style/translation) · [Accessibility](https://developers.google.com/style/accessibility). Verify a disputed or missing rule against the live page → `references/style-sources.md`.

Upstream: [Articles (a, an, the)](https://developers.google.com/style/articles) · [Pronouns](https://developers.google.com/style/pronouns) · [Possessives](https://developers.google.com/style/possessives) · [Pluralization](https://developers

Upstream: [Figures and other images](https://developers.google.com/style/images). Verify a disputed or missing rule against the live page → `references/style-sources.md`.

Upstream: [Inclusive language](https://developers.google.com/style/inclusive-documentation). Verify a disputed or missing rule against the live page → `references/style-sources.md`.

Load for any wording, formatting, or terminology question. This maps every topic in the [Google developer documentation style guide](https://developers.google.com/style) to the reference that owns its rules, so the guide works as a lookup i

Upstream: [Highlights](https://developers.google.com/style/highlights) · [Philosophy of this guide](https://developers.google.com/style/philosophy) · [What's new](https://developers.google.com/style/whats-new). Verify a disputed or missing 

Upstream: [Cross-references and linking](https://developers.google.com/style/cross-references) · [Headings as link targets](https://developers.google.com/style/headings-targets). Verify a disputed or missing rule against the live page → `re

Upstream: [Numbers](https://developers.google.com/style/numbers) · [Dates and times](https://developers.google.com/style/dates-times) · [Units of measurement](https://developers.google.com/style/units-of-measure) · [Phone numbers](https://d

Upstream: [Procedures](https://developers.google.com/style/procedures). Verify a disputed or missing rule against the live page → `references/style-sources.md`.

Upstream: [Colons](https://developers.google.com/style/colons) · [Commas](https://developers.google.com/style/commas) · [Dashes](https://developers.google.com/style/dashes) · [Ellipses](https://developers.google.com/style/ellipses) · [Hyphe

Upstream: [Highlights](https://developers.google.com/style/highlights) · [Google developer documentation style guide](https://developers.google.com/style/). Verify a disputed or missing rule against the live page → `references/style-sources

Load when someone disputes a rule, when the pack doesn't carry the rule you need, or to check whether the guide has moved. Why: these references are a snapshot, and the live guide is the authority. Every page sits at `https://developers.goo

| Page | Open it for | |---|---| | [Google developer documentation style guide](https://developers.google.com/style) | The guide home and its own navigation |

|---|---| | [Google developer documentation style guide](https://developers.google.com/style) | The guide home and its own navigation | | [Highlights](https://developers.google.com/style/highlights) | The short list of rules that cover most

| [Google developer documentation style guide](https://developers.google.com/style) | The guide home and its own navigation |

| [Highlights](https://developers.google.com/style/highlights) | The short list of rules that cover most reviews |

| [What's new](https://developers.google.com/style/whats-new) | The changelog — read before calling a rule here stale |

| [Philosophy of this guide](https://developers.google.com/style/philosophy) | What the guide optimizes for when two rules pull apart |

| [Word list](https://developers.google.com/style/word-list) | The source of `assets/google-word-list.tsv` |

| [Google site policies](https://developers.google.com/site-policies) | The [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/) terms these references restate under |

Upstream: [Headings and titles](https://developers.google.com/style/headings) · [Lists](https://developers.google.com/style/lists). Verify a disputed or missing rule against the live page → `references/style-sources.md`.

Upstream: [UI elements and interaction](https://developers.google.com/style/ui-elements). Verify a disputed or missing rule against the live page → `references/style-sources.md`.

Upstream: [Voice and tone](https://developers.google.com/style/tone) · [Second person](https://developers.google.com/style/person) · [Active voice](https://developers.google.com/style/voice) · [Present tense](https://developers.google.com/s

Upstream: [Word list](https://developers.google.com/style/word-list). Verify a disputed or missing rule against the live page → `references/style-sources.md`.

#!/usr/bin/env node

--json      machine-readable summary   --out       target TSV (default assets/google-word-list.tsv)   --url       source page (default https://developers.google.com/style/word-list)`);

--out       target TSV (default assets/google-word-list.tsv)   --url       source page (default https://developers.google.com/style/word-list)`);   process.exit(0);

--url       source page (default https://developers.google.com/style/word-list)`);

const url = val('--url', 'https://developers.google.com/style/word-list');

let html; try {   const res = await fetch(url, { headers: { 'user-agent': 'octocode-documentation/refresh-word-list' } });

try {   const res = await fetch(url, { headers: { 'user-agent': 'octocode-documentation/refresh-word-list' } });   if (!res.ok) throw new Error(`HTTP ${res.status}`);

const res = await fetch(url, { headers: { 'user-agent': 'octocode-documentation/refresh-word-list' } });

#!/usr/bin/env node

To learn more, click [here](https://example.com) or see [https://example.com/docs](https://example.com/docs).

For more information about retention windows, see [Retention policy reference](https://example.com/retention).

[Retention policy reference](https://example.com/retention).

#!/usr/bin/env node

#!/usr/bin/env node

#!/usr/bin/env node

The generated page is an **agent control surface**: it registers 11 page-native MCP tools via `document.modelContext.registerTool` ([WebMCP](https://webmcp.dev/)). An agent launches Chrome with

`document.modelContext.registerTool` ([WebMCP](https://webmcp.dev/)). An agent launches Chrome with

{   "imports": {     "three": "https://unpkg.com/three@latest/build/three.module.js",

"imports": {     "three": "https://unpkg.com/three@latest/build/three.module.js",     "three/addons/": "https://unpkg.com/three@latest/examples/jsm/"

"three": "https://unpkg.com/three@latest/build/three.module.js",

"three/addons/": "https://unpkg.com/three@latest/examples/jsm/"

cervical/thoracic/lumbar column (24 vertebrae) and carpal/tarsal/phalangeal clusters (54 small bones) into rig-practical groups, matching the Unity Humanoid (15 required bones) / VRM 1.0 Humanoid (13 required, ~56 total) convention rather t

convention rather than modeling all 206 bones. [VRM humanoid spec](https://github.com/vrm-c/vrm-specification/blob/master/specification/VRMC_vrm-1.0/humanoid.md) · [OpenStax skeletal divisions](https://openstax.org/books/anatomy-and-physiol

as an independent box limit. Swing-twist is the correct model for shoulder/hip because it bounds cone + twist separately, matching physics-engine `ConeTwistConstraint` (ammo.js/cannon-es); box clamps are what `CCDIKSolver`/`three-ik` use an

`CCDIKSolver`/`three-ik` use and are adequate for the smaller-range joints below. [Daniel Holden, Joint Limits](https://theorangeduck.com/page/joint-limits) · [cannon-es ConeTwistConstraint](https://pmndrs.github.io/cannon-es/docs/classes/C

| Source | Used for | |---|---| | [OpenStax A&P 2e](https://openstax.org/books/anatomy-and-physiology-2e/pages/7-1-divisions-of-the-skeletal-system) | Bone divisions |

|---|---| | [OpenStax A&P 2e](https://openstax.org/books/anatomy-and-physiology-2e/pages/7-1-divisions-of-the-skeletal-system) | Bone divisions | | [NCBI Appendicular Skeleton](https://www.ncbi.nlm.nih.gov/books/NBK535397/) | Limb bone chai

| [OpenStax A&P 2e](https://openstax.org/books/anatomy-and-physiology-2e/pages/7-1-divisions-of-the-skeletal-system) | Bone divisions |

| [NCBI Appendicular Skeleton](https://www.ncbi.nlm.nih.gov/books/NBK535397/) | Limb bone chain |

| [NCBI Glenohumeral](https://www.ncbi.nlm.nih.gov/books/NBK537018/) / [TMJ](https://www.ncbi.nlm.nih.gov/books/NBK538486/) | Shoulder / jaw DOF |

| [goniometer.io](https://goniometer.io/range-of-motion) (AAOS) | Primary ROM table; 6 corrections |

| [OrthoFixar](https://orthofixar.com/) | Independent ROM cross-check |

| [VRM 1.0 humanoid](https://github.com/vrm-c/vrm-specification/blob/master/specification/VRMC_vrm-1.0/humanoid.md) / [Unity Avatar](https://docs.unity3d.com/Manual/ConfiguringtheAvatar.html) | Rig bone count; rest pose |

| Source | Used for | |---|---| | [three.js Skeleton](https://threejs.org/docs/#api/en/objects/Skeleton) / [Skeleton.js](https://github.com/mrdoob/three.js/blob/dev/src/objects/Skeleton.js) | FK hierarchy |

|---|---| | [three.js Skeleton](https://threejs.org/docs/#api/en/objects/Skeleton) / [Skeleton.js](https://github.com/mrdoob/three.js/blob/dev/src/objects/Skeleton.js) | FK hierarchy | | [CCDIKSolver](https://threejs.org/docs/#examples/en/a

| [three.js Skeleton](https://threejs.org/docs/#api/en/objects/Skeleton) / [Skeleton.js](https://github.com/mrdoob/three.js/blob/dev/src/objects/Skeleton.js) | FK hierarchy |

| [CCDIKSolver](https://threejs.org/docs/#examples/en/animations/CCDIKSolver) / [THREE.IK](https://github.com/jsantell/THREE.IK) | IK shape (stale FABRIK ref) |

| [Holden joint limits](https://theorangeduck.com/page/joint-limits) | Swing-twist |

| [Matt Rossman bone orientations](https://mattrossman.com/2024/07/10/visualizing-threejs-bone-orientations/) | Why SkeletonHelper alone fails |

| Source | Used for | |---|---| | [AnimSchool run cycle](https://blog.animschool.edu/2024/04/10/the-key-poses-of-a-run-cycle/) / [MoCap Online](https://mocaponline.com/blogs/mocap-news/run-cycle-animation) | Contact/flight; root motion |

|---|---| | [AnimSchool run cycle](https://blog.animschool.edu/2024/04/10/the-key-poses-of-a-run-cycle/) / [MoCap Online](https://mocaponline.com/blogs/mocap-news/run-cycle-animation) | Contact/flight; root motion | | [animcoding blend](htt

| [AnimSchool run cycle](https://blog.animschool.edu/2024/04/10/the-key-poses-of-a-run-cycle/) / [MoCap Online](https://mocaponline.com/blogs/mocap-news/run-cycle-animation) | Contact/flight; root motion |

| [animcoding blend](https://animcoding.com/post/animation-tech-intro-part-3-blending/) / [Yuksel Catmull-Rom](https://www.cemyuksel.com/research/catmullrom_param/catmullrom.pdf) | Ease/slerp; centripetal α=0.5 |

| [PMC gait CoM](https://pmc.ncbi.nlm.nih.gov/articles/PMC6763727/) / Unreal foot-IK pelvis offset | Grounding bob |

| [gameanim 12 principles](https://www.gameanim.com/2019/05/15/the-12-principles-of-animation-in-video-games/) / [Animator Island music](https://www.animatorisland.com/animating-to-music/) | Landing settle; dance beat → `assets/` |

| Source | Used for | |---|---| | [ASU robotics control](https://wanxinjin.github.io/asu-robotics/lec19/control_overview.html) | Joint- vs task-space; FK |

|---|---| | [ASU robotics control](https://wanxinjin.github.io/asu-robotics/lec19/control_overview.html) | Joint- vs task-space; FK | | [Unity IK](https://docs.unity3d.com/Manual/InverseKinematics.html) / [MoveIt practices](https://docs.pic

| [ASU robotics control](https://wanxinjin.github.io/asu-robotics/lec19/control_overview.html) | Joint- vs task-space; FK |

| [Unity IK](https://docs.unity3d.com/Manual/InverseKinematics.html) / [MoveIt practices](https://docs.picknik.ai/how_to/robotics_applications/motion_planning_best_practices/) | `reach`/`look_at`; report clamp |

| [WebMCP](https://webmcp.dev/) / [webmcp repo](https://github.com/webmachinelearning/webmcp) | Page-side tool API |

`scripts/skeleton.mjs`. Values are clinical/kinesiology population norms (AAOS *Joint Motion* chart cross-checked against independent sources) — **rig defaults, not hard biomechanical limits**; real individuals vary. Two research passes val

against independent sources) — **rig defaults, not hard biomechanical limits**; real individuals vary. Two research passes validated these against [goniometer.io](https://goniometer.io/range-of-motion), [OrthoFixar](https://orthofixar.com/)

research passes validated these against [goniometer.io](https://goniometer.io/range-of-motion),

[OrthoFixar](https://orthofixar.com/), [Kenhub](https://www.kenhub.com/en/library/anatomy), Physiopedia, and

Load when an agent should drive the running manikin as a live MCP server instead of editing pose files. The viewer registers page-native tools via `document.modelContext.registerTool` (the WebMCP standard — [webmcp.dev](https://webmcp.dev/)

viewer registers page-native tools via `document.modelContext.registerTool` (the WebMCP standard — [webmcp.dev](https://webmcp.dev/), [webmachinelearning/webmcp](https://github.com/webmachinelearning/webmcp)); an agent discovers and invokes

[webmcp.dev](https://webmcp.dev/), [webmachinelearning/webmcp](https://github.com/webmachinelearning/webmcp));

#!/usr/bin/env node

## Sources - A2A, [Protocol specification](https://github.com/a2aproject/A2A/blob/main/docs/specification.md) — Agent Cards, Tasks, Messages, Artifacts, capability checks, authorization, and validation.

- A2A, [Protocol specification](https://github.com/a2aproject/A2A/blob/main/docs/specification.md) — Agent Cards, Tasks, Messages, Artifacts, capability checks, authorization, and validation.

- Model Context Protocol, [Tasks](https://modelcontextprotocol.io/specification/2025-11-25/basic/utilities/tasks) — task lifecycle support is experimental in the 2025-11-25 specification.

- OpenAI Agents SDK, [composition patterns](https://openai.github.io/openai-agents-js/guides/agents/) and [handoffs](https://openai.github.io/openai-agents-js/guides/handoffs/) — manager versus ownership transfer and filtered/typed handoff 

## Sources - OpenAI, [Prompt engineering](https://developers.openai.com/api/docs/guides/prompt-engineering) — Markdown/XML boundaries and developer-message structure.

- OpenAI, [Prompt engineering](https://developers.openai.com/api/docs/guides/prompt-engineering) — Markdown/XML boundaries and developer-message structure.

- Anthropic, [Effective context engineering for AI agents](https://www.anthropic.com/engineering/effective-context-engineering-for-ai-agents) — bounded, high-signal context for agents.

- Liu et al., [Lost in the Middle](https://arxiv.org/abs/2307.03172) — context position can affect long-context retrieval; validate placement on the target task.

- Xiao et al., [StreamingLLM](https://arxiv.org/abs/2309.17453) — attention sinks are an internal model/inference mechanism.

## Sources - Anthropic, [Effective context engineering for AI agents](https://www.anthropic.com/engineering/effective-context-engineering-for-ai-agents) — the minimal high-signal context principle.

- Anthropic, [Effective context engineering for AI agents](https://www.anthropic.com/engineering/effective-context-engineering-for-ai-agents) — the minimal high-signal context principle.

## Sources - Anthropic, [Effective context engineering for AI agents](https://www.anthropic.com/engineering/effective-context-engineering-for-ai-agents) — finite attention budget and minimal high-signal context.

- Anthropic, [Effective context engineering for AI agents](https://www.anthropic.com/engineering/effective-context-engineering-for-ai-agents) — finite attention budget and minimal high-signal context.

- Anthropic, [Writing effective tools for AI agents](https://www.anthropic.com/engineering/writing-tools-for-agents) — filtering, pagination, truncation, and useful continuation guidance.

## Sources - Anthropic, [Writing effective tools for AI agents](https://www.anthropic.com/engineering/writing-tools-for-agents) — realistic evaluation tasks, verifiers, held-out tests, and tool-use metrics.

- Anthropic, [Writing effective tools for AI agents](https://www.anthropic.com/engineering/writing-tools-for-agents) — realistic evaluation tasks, verifiers, held-out tests, and tool-use metrics.

- Anthropic, [Effective context engineering for AI agents](https://www.anthropic.com/engineering/effective-context-engineering-for-ai-agents) — evaluate context quality as part of agent behavior.

## Sources - Anthropic, [Effective context engineering for AI agents](https://www.anthropic.com/engineering/effective-context-engineering-for-ai-agents) — clear, specific prompts at the right level of prescription.

- Anthropic, [Effective context engineering for AI agents](https://www.anthropic.com/engineering/effective-context-engineering-for-ai-agents) — clear, specific prompts at the right level of prescription.

## Sources - Anthropic, [Effective context engineering for AI agents](https://www.anthropic.com/engineering/effective-context-engineering-for-ai-agents) — start minimal, then add instructions from observed failure modes.

- Anthropic, [Effective context engineering for AI agents](https://www.anthropic.com/engineering/effective-context-engineering-for-ai-agents) — start minimal, then add instructions from observed failure modes.

## Sources - Model Context Protocol, [Tools specification](https://modelcontextprotocol.io/specification/2025-06-18/server/tools) — explicit output structure and error signaling support reliable tool use.

- Model Context Protocol, [Tools specification](https://modelcontextprotocol.io/specification/2025-06-18/server/tools) — explicit output structure and error signaling support reliable tool use.

## Sources - Anthropic, [Effective context engineering for AI agents](https://www.anthropic.com/engineering/effective-context-engineering-for-ai-agents) — distinct prompt sections and a specific-but-not-brittle instruction altitude.

- Anthropic, [Effective context engineering for AI agents](https://www.anthropic.com/engineering/effective-context-engineering-for-ai-agents) — distinct prompt sections and a specific-but-not-brittle instruction altitude.

## Sources - OpenAI, [Prompt caching](https://developers.openai.com/api/docs/guides/prompt-caching) — exact-prefix layout, automatic caching, cache keys/retention, and `cached_tokens` telemetry.

- OpenAI, [Prompt caching](https://developers.openai.com/api/docs/guides/prompt-caching) — exact-prefix layout, automatic caching, cache keys/retention, and `cached_tokens` telemetry.

- Anthropic, [Prompt caching](https://platform.claude.com/docs/en/build-with-claude/prompt-caching) — prefix breakpoints, automatic caching, TTLs, and cache usage fields.

- Anthropic, [Tool use with prompt caching](https://platform.claude.com/docs/en/agents-and-tools/tool-use/tool-use-with-prompt-caching) — stable tool definitions and cache invalidation considerations.

## Sources - DAIR.AI, [Prompt Engineering Guide](https://github.com/dair-ai/Prompt-Engineering-Guide) — broad technique landscape, papers, and references.

- DAIR.AI, [Prompt Engineering Guide](https://github.com/dair-ai/Prompt-Engineering-Guide) — broad technique landscape, papers, and references.

- OpenAI, [Prompt engineering](https://developers.openai.com/api/docs/guides/prompt-engineering) — structured sections, examples, retrieval context, tool guidance, model-specific prompting, and evaluation.

- Anthropic, [Prompt engineering overview](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/overview) and [context engineering](https://www.anthropic.com/engineering/effective-context-engineering-for-ai-agents) — suc

## Sources - Anthropic, [Writing effective tools for AI agents](https://www.anthropic.com/engineering/writing-tools-for-agents) — evaluate real tasks, inspect transcripts, and measure tool behavior.

- Anthropic, [Writing effective tools for AI agents](https://www.anthropic.com/engineering/writing-tools-for-agents) — evaluate real tasks, inspect transcripts, and measure tool behavior.

## Sources - Anthropic, [Writing effective tools for AI agents](https://www.anthropic.com/engineering/writing-tools-for-agents) — namespacing, clear schemas, response formats, and token-efficient results.

- Anthropic, [Writing effective tools for AI agents](https://www.anthropic.com/engineering/writing-tools-for-agents) — namespacing, clear schemas, response formats, and token-efficient results.

- Model Context Protocol, [Tools specification](https://modelcontextprotocol.io/specification/2025-06-18/server/tools) — tool metadata, input/output schemas, and paginated discovery.

## Sources - Model Context Protocol, [Tools specification](https://modelcontextprotocol.io/specification/2025-06-18/server/tools) — tool annotations are untrusted unless supplied by a trusted server.

- Model Context Protocol, [Tools specification](https://modelcontextprotocol.io/specification/2025-06-18/server/tools) — tool annotations are untrusted unless supplied by a trusted server.

- Anthropic, [Effective context engineering for AI agents](https://www.anthropic.com/engineering/effective-context-engineering-for-ai-agents) — separate clear instructions from external context and retain only high-signal information.

## Sources - Anthropic, [Writing effective tools for AI agents](https://www.anthropic.com/engineering/writing-tools-for-agents) — use verifiable outcomes and measure errors, calls, and tokens alongside task success.

- Anthropic, [Writing effective tools for AI agents](https://www.anthropic.com/engineering/writing-tools-for-agents) — use verifiable outcomes and measure errors, calls, and tokens alongside task success.

## Sources - Zod, [Defining schemas](https://zod.dev/api) — discriminated unions and type narrowing.

- Zod, [Defining schemas](https://zod.dev/api) — discriminated unions and type narrowing.

- Zod, [JSON Schema](https://zod.dev/json-schema) — stable `z.toJSONSchema()` conversion; `z.fromJSONSchema()` is experimental.

- A2A, [Protocol specification](https://github.com/a2aproject/A2A/blob/main/docs/specification.md) — required-field validation, schema validation, authorization, and injection protections.

{"repo":"owner/name","url":"https://github.com/owner/name","sourceQuery":"term","stars":1200,"language":"TypeScript","license":"MIT","lastActivity":"2026-06-01","package":"name","fit":"high","activity":"active","evidenceIds":["ev1"],"notes"

#!/usr/bin/env node

1. If `octocode-research` is installed, load it and request the needed RFC evidence. 2. If it is not installed, point the user or agent to https://github.com/bgauryy/octocode/tree/main/skills/octocode-research.

2. If it is not installed, point the user or agent to https://github.com/bgauryy/octocode/tree/main/skills/octocode-research.

Use `octocode-research` if installed. If it is missing, use https://github.com/bgauryy/octocode/tree/main/skills/octocode-research or install it with:

Public pages → local cited corpus. Live clicks/HAR/perf → [`octocode-chrome-devtools`](https://github.com/bgauryy/octocode/tree/main/skills/octocode-chrome-devtools).

[Credential-related source omitted]

Use [ScrapingAnt](https://scrapingant.com/?ref=mty5mzy) when keyless routes fail and the user approves credits (anti-bot, markdown, AI extraction).

# Get a key: https://scrapingant.com/?ref=mty5mzy

[Credential-related source omitted]

[Credential-related source omitted]

# Explicit hosted fetch (not auto):

node skills/octocode-scraping/scripts/fetch.mjs --provider scrapingant --url 'https://example.com'

Full config: [CONFIGURATION.md](https://github.com/bgauryy/octocode/blob/main/docs/CONFIGURATION.md).

Skill scripts run standalone. For local corpus search tools in the IDE: `npx octocode install --ide cursor` — [OCTOCODE_MCP.md](https://github.com/bgauryy/octocode/blob/main/docs/OCTOCODE_MCP.md).

## Contract `fetch({ url, pageId, config, apiKey })` → `FetchResponse` (`scripts/schemas/provider.schema.json`). Corpus/analyzers must not branch on vendor names.

`fetch({ url, pageId, config, apiKey })` → `FetchResponse` (`scripts/schemas/provider.schema.json`). Corpus/analyzers must not branch on vendor names.

node skills/octocode-scraping/scripts/provider-check.mjs --provider scrapingant node skills/octocode-scraping/scripts/fetch.mjs --url https://example.com --provider scrapingant --mode html

node skills/octocode-scraping/scripts/fetch.mjs --url https://example.com --provider scrapingant --mode html

node skills/octocode-scraping/scripts/fetch.mjs --url https://example.com --provider scrapingant --mode markdown

node skills/octocode-scraping/scripts/fetch.mjs --url https://example.com --provider scrapingant --mode extract --extract-properties "title, content"

#!/usr/bin/env node

#!/usr/bin/env node

#!/usr/bin/env node

#!/usr/bin/env node

#!/usr/bin/env node

#!/usr/bin/env node

pageIndex += 1;   const pageId = `page-${String(pageIndex).padStart(3, '0')}`;   const response = await provider.fetch({ url, pageId, config, apiKey });

const pageId = `page-${String(pageIndex).padStart(3, '0')}`;   const response = await provider.fetch({ url, pageId, config, apiKey });   const written = await writePage({ sessionDir, config, response, pageIndex });

const response = await provider.fetch({ url, pageId, config, apiKey });

#!/usr/bin/env node

#!/usr/bin/env node

Executable file

Executable file

Executable file

Executable file

export async function fetchScrapingAnt({ url, pageId, config, apiKey }) {   const apiUrl = new URL(`https://api.scrapingant.com/v2/${config.endpoint}`);

const apiUrl = new URL(`https://api.scrapingant.com/v2/${config.endpoint}`);

creditCost = config.mockCreditCost;     } else {       const res = await fetch(apiUrl, { headers: { 'user-agent': 'octocode-scraping/0.1' } });

} else {       const res = await fetch(apiUrl, { headers: { 'user-agent': 'octocode-scraping/0.1' } });       status = res.status;

const res = await fetch(apiUrl, { headers: { 'user-agent': 'octocode-scraping/0.1' } });

body = config.mockBodyFile ? await readFile(config.mockBodyFile, 'utf8') : '';     } else {       const res = await fetch(url, { headers: DIRECT_STEALTH_HEADERS });

} else {       const res = await fetch(url, { headers: DIRECT_STEALTH_HEADERS });       status = res.status;

const res = await fetch(url, { headers: DIRECT_STEALTH_HEADERS });

try {     const smUrl = new URL('/sitemap.xml', targetUrl).href;     const sm = await fetch(smUrl).then((r) => r.text());

const smUrl = new URL('/sitemap.xml', targetUrl).href;     const sm = await fetch(smUrl).then((r) => r.text());     const locs = [...sm.matchAll(/<loc>(.*?)<\/loc>/g)].map((m) => m[1].trim()).filter(Boolean);

const sm = await fetch(smUrl).then((r) => r.text());

Executable file

await writeFile(join(sessionDir, 'reports/summary.md'), `# Scrape Summary\n\nTarget: ${config.targetUrl}\nRoute: ${config.provider}:${config.mode}\nFetched: ${new Date().toISOString()}\nPages: ${sources.length}\nStatus: ${first.status || 'f

Executable file

Executable file

Executable file

#!/usr/bin/env node

#!/usr/bin/env node

const url = new URL('https://api.scrapingant.com/v2/usage'); url.searchParams.set('x-api-key', apiKey); const res = await fetch(url, { headers: { 'user-agent': 'octocode-scraping-usage/0.1' } });

url.searchParams.set('x-api-key', apiKey); const res = await fetch(url, { headers: { 'user-agent': 'octocode-scraping-usage/0.1' } }); const text = await res.text();

const res = await fetch(url, { headers: { 'user-agent': 'octocode-scraping-usage/0.1' } });

#!/usr/bin/env node

#!/usr/bin/env node

{   "$schema": "https://json-schema.org/draft/2020-12/schema",

"$schema": "https://json-schema.org/draft/2020-12/schema",

"$id": "https://octocode.dev/schemas/scraping/graph.schema.json",

{   "$schema": "https://json-schema.org/draft/2020-12/schema",

"$schema": "https://json-schema.org/draft/2020-12/schema",

"$id": "https://octocode.dev/schemas/scraping/provider.schema.json",

"FetchResponse": {       "type": "object",       "description": "Return shape every provider.fetch({ url, pageId, config, apiKey }) must produce.",

"type": "object",       "description": "Return shape every provider.fetch({ url, pageId, config, apiKey }) must produce.",       "required": ["pageId", "url", "status", "contentType", "body", "fetchError", "creditCost", "fetchedAt"],

"description": "Return shape every provider.fetch({ url, pageId, config, apiKey }) must produce.",

#!/usr/bin/env node

#!/usr/bin/env node

#!/usr/bin/env node

#!/usr/bin/env node

#!/usr/bin/env bash

| Keyword/topic search | skills.sh `/api/search?q=` (sort `installs`) ∥ GitHub ∥ web | | Mentions outside registries | WebSearch → confirm `SKILL.md` via Octocode | | Is X published / battle-tested? | `https://www.skills.sh/<owner>/<repo>/<

| Mentions outside registries | WebSearch → confirm `SKILL.md` via Octocode | | Is X published / battle-tested? | `https://www.skills.sh/<owner>/<repo>/<skill>` | | Most-installed now | `https://www.skills.sh` leaderboard |

| Is X published / battle-tested? | `https://www.skills.sh/<owner>/<repo>/<skill>` |

| Most-installed now | `https://www.skills.sh` leaderboard |

- packages: `octocode tools npmSearch --queries '{"packageName":"<package>"}' --compact`    Read `octocode tools <name> --scheme` before any raw tool call. 3. ELSE point to https://github.com/bgauryy/octocode/tree/main/skills/octocode-resea

3. ELSE point to https://github.com/bgauryy/octocode/tree/main/skills/octocode-research. Install only with user approval: `npx octocode skill --name octocode-research` (add `--platform <host>` for a specific host).

- **Install count** — `skills.sh/api/search?q=` (curl in `search-playbook.md`); sort by `installs`. High installs + modest stars usually beats the reverse.

curl 'https://www.skills.sh/api/search?q={{SEARCH_KEY}}&limit=100' --compressed \

- Rank by installs/recency/audits → `references/quality-signals.md`. - Registry/CLI catalog → `references/discovery-surfaces.md`. - External index (when web allowed): `https://agentskills.io/llms.txt`.

- External index (when web allowed): `https://agentskills.io/llms.txt`.

#!/usr/bin/env bash

#!/usr/bin/env node

#!/usr/bin/env node

#!/usr/bin/env node

Spec: https://a2a-protocol.org/latest/specification/

**Not required.** Do not treat tags below as must-install. Always start from live `ollama list`. Portable routing lives in `references/model-selection.md`; catalog / RAM kits in `references/ollama-local-models.md`. Sources for examples: [ge

1. If `octocode-research` is installed, load it inside the researcher/architect worker (or parent) for the probe. 2. If missing, point to https://github.com/bgauryy/octocode/tree/main/skills/octocode-research

2. If missing, point to https://github.com/bgauryy/octocode/tree/main/skills/octocode-research

curl -sS "$OLLAMA_HOST/api/tags"           # ≈ list

curl -sS "$OLLAMA_HOST/api/ps"             # ≈ ps

curl -sS "$OLLAMA_HOST/api/generate" -d '{...}'

curl -sS "$OLLAMA_HOST/api/chat" -d '{...}'

Commands and flags verified against local `ollama --help` / subcommand help (Ollama CLI on the agent host). Official docs: https://github.com/ollama/ollama · library hub: https://ollama.com/library · key families: [gemma4](https://ollama.co

| `ollama rm MODEL` | Delete a local model — **ask user** |

**Prerequisites:** Ollama installed and reachable (`ollama serve` or OS service); at least one **chat-capable** model installed (`ollama list`); selected model exported as `OLLAMA_WORKER_MODEL` (exact name from list); `OLLAMA_HOST` default 

**Health check** — daemon only: `./scripts/ollama-health.sh` · after ROUTE (model selected): `./scripts/ollama-health.sh --model "$OLLAMA_WORKER_MODEL"` · manual: `curl -sS "${OLLAMA_HOST:-http://127.0.0.1:11434}/api/tags"` or `ollama list`

Legend: **Y** = library tag and/or `ollama show`; **—** = not claimed; **?** = verify after pull. Sizes/context from [ollama.com/library](https://ollama.com/library) (re-check with `ollama show <tag>`). **Thinking vs reasoning:** `thinking`

| Claim | Source | Confidence | |-------|--------|------------| | Gemma 4 vs Qwen 3.5 tradeoffs (context, multilingual, size ladder) | [MindStudio comparison](https://www.mindstudio.ai/blog/gemma-4-vs-qwen-3-5-open-weight-comparison) (2026-

|-------|--------|------------| | Gemma 4 vs Qwen 3.5 tradeoffs (context, multilingual, size ladder) | [MindStudio comparison](https://www.mindstudio.ai/blog/gemma-4-vs-qwen-3-5-open-weight-comparison) (2026-04) | Medium | | Qwen often pref

| Gemma 4 vs Qwen 3.5 tradeoffs (context, multilingual, size ladder) | [MindStudio comparison](https://www.mindstudio.ai/blog/gemma-4-vs-qwen-3-5-open-weight-comparison) (2026-04) | Medium |

| Qwen often preferred for agents; Gemma for efficient local | [Codersera 2026 comparison](https://codersera.com/blog/gemma-4-vs-qwen-3-5-comparison-2026/) | Medium |

| Local ABS bench: `qwen3.5:9b` top overall; all ≥7B saturate tool pick | [JConradoN/local-llm-benchmark](https://github.com/JConradoN/local-llm-benchmark) | Medium |

| Gemma 4 31B strong LiveCodeBench / Codeforces vs prior Gemma | Ollama Gemma 4 readme + [ai.rs writeup](https://ai.rs/ai-developer/gemma-4-vs-qwen-3-5-vs-llama-4-compared) | Medium (benchmarks move) |

| Gemma 4 31B often more token-efficient than Qwen3.5 27B when thinking | [Kaitchup Substack](https://kaitchup.substack.com/p/gemma-4-31b-vs-qwen35-27b-inference) | Medium |

- https://ollama.com/library — catalog popularity and families

- https://ollama.com/library/gemma4 — sizes, ctx, tools/thinking/vision/audio, benchmarks

- https://ollama.com/library/qwen3.5 — size ladder, 256K, multimodal, tools/thinking

- https://ollama.com/library/qwen3.6 — agentic coding / thinking preservation

- https://ollama.com/library/qwen3-coder — 30B MoE coding agent

- https://ollama.com/library/gpt-oss — tools, thinking, agent features

- https://ollama.com/library/north-mini-code-1.0 — SWE agent MoE, long context tags

- https://ollama.com/library/laguna-xs-2.1 — local long-horizon coding MoE

- https://ollama.com/library/lfm2.5 — edge tool calling

- https://ollama.com/library/deepseek-r1 — reasoning family

- https://ollama.com/library/qwen2.5 — legacy general Qwen (32K)

- https://ollama.com/library/qwen2.5-coder — legacy coder sizes / 32K

- https://ollama.com/library/devstral — older SWE agent

- https://ollama.com/library/codestral — FIM coder (no tools tag)

- https://ollama.com/library/granite4.1 — 3b/8b/30b, 128K, tools

- https://ollama.com/library/glm-ocr — OCR, vision+tools, ~2.2 GB / 128K

- https://ollama.com/search?c=tools — tools-capable set

- https://ollama.com/search?c=thinking — thinking-capable set

- https://ollama.com/blog — MLX / Gemma 4 coding-agent speed (2026-06)

- https://www.mindstudio.ai/blog/gemma-4-vs-qwen-3-5-open-weight-comparison

- https://codersera.com/blog/gemma-4-vs-qwen-3-5-comparison-2026/

- https://github.com/JConradoN/local-llm-benchmark

- https://ai.rs/ai-developer/gemma-4-vs-qwen-3-5-vs-llama-4-compared

- https://kaitchup.substack.com/p/gemma-4-31b-vs-qwen35-27b-inference

#!/usr/bin/env bash

HOST="${OLLAMA_HOST:-http://127.0.0.1:11434}"

if ! command -v curl >/dev/null 2>&1; then

echo "error: curl is required" >&2

if ! curl -sS --max-time 3 "${HOST}/api/tags" >/dev/null; then

echo "error: Ollama not reachable at ${HOST}" >&2

TAGS="$(curl -sS --max-time 5 "${HOST}/api/tags")"

echo "ok: ollama up at ${HOST}"

#!/usr/bin/env bash

HOST="${OLLAMA_HOST:-http://127.0.0.1:11434}"

Env:   OLLAMA_HOST              default http://127.0.0.1:11434

OLLAMA_HOST              default http://127.0.0.1:11434

cleanup() { rm -f "${PROMPT_FILE}"; }

')"   local resp   resp="$(curl -sS -X POST "${HOST}/api/generate" \

local resp   resp="$(curl -sS -X POST "${HOST}/api/generate" \     -H "Content-Type: application/json" \

resp="$(curl -sS -X POST "${HOST}/api/generate" \

-H "Content-Type: application/json" \     -d "${payload}")" || {     echo "error: HTTP generate failed against ${HOST}" >&2

-d "${payload}")" || {     echo "error: HTTP generate failed against ${HOST}" >&2     exit 1

echo "error: HTTP generate failed against ${HOST}" >&2