id	readme
title	Pine Script v6 knowledge base
type	index
group	root
tags	index pine-v6
pine_version	v6

Pine Script v6 knowledge base

This knowledge base consolidates the official "Pine Script® User Manual" and "Pine Script® Language Reference Manual" for version 6, structured for both human developers and Large Language Models (LLMs).

Deterministic, offline, and optimized for both human developers and Large Language Models (LLMs): 941 entities, 128 documents, 57 namespace digests, zero broken internal links, one validator you can run yourself.

Table of contents

• Objective
• Common use cases
• Repository map
• Three ways to read a namespace
• Navigation contract
• Largest namespaces
• For language models and agents
• Verify it yourself
• Provenance and limits
• Tooling
• Contributing
• Maintainer and contact
• License and attribution
• Audit trail

Objective

Turn a raw documentation dump into a knowledge base where a human finds the syntax to fix an error in fewer than three steps, and a language model answers without guessing. Every path, identifier and anchor is derived from a canonical name, so two builds of the same input produce the same tree.

Common use cases

I need to...	Go here	Steps
Fix an error code	errors/_index.md	2
Check one signature	INDEX.md	2
Write code in a namespace	reference/<ns>/_index.md	2
Migrate a v5 script	migration/to-pine-version-6.md	1
Copy working code	examples/_index.md	2
Wire this into an agent	AGENTS.md	1

Repository map

Path	Contents
guide/	Practical guide: primer, language, visuals, concepts, writing.
reference/	941 entities in 57 namespace folders; each folder opens with a digest.
migration/	v5 to v6 breaking changes and release notes.
examples/	Runnable pair for most guide pages.
errors/	Error and warning codes with fixes.
faq/	Recurring questions.
dist/bundles/	Generated single-file bundle per namespace.
.llm/	Machine layer: JSONL records, schemas, resolver, chunks.
scripts/validate.py	Invariant checker, standard library only.

Three ways to read a namespace

Route	What you get	Tokens for array
Digest reference/array/_index.md	every signature, parameter and return type	about 2273
Member page reference/array/get.md	one entity in full, with remarks and example	about 420
Bundle dist/bundles/array.md	all member pages concatenated	about 20825

Start with the digest. Open a member page for detail. Use the bundle only when you need most of the namespace at once.

Navigation contract

• Maximum depth is three levels.
• Every folder exposes _index.md as its entry point.
• Names are deterministic: ta.sma() is always reference/ta/sma.md.
• Entities sharing a prefix live together, so reference/ta/ is the whole ta namespace.
• Generated folders are dist/, .llm/ and the three root indexes. Never edit them by hand.

Largest namespaces

Namespace	Entities	Digest	Bundle
strategy	96	4105 tokens	41308 tokens
ta	69	2870 tokens	21699 tokens
currency	56	1163 tokens	6395 tokens
array	55	2273 tokens	20825 tokens
matrix	49	2015 tokens	20999 tokens
label	43	1773 tokens	12451 tokens
syminfo	42	1424 tokens	18862 tokens
box	30	1498 tokens	8381 tokens

For language models and agents

Artifact	Answers
llms.txt	Where do I start?
AGENTS.md	What rules must I obey when generating code?
.llm/entrypoints.json	Which file answers my task at the lowest cost?
.llm/context-packs.json	What fits in my context window?
.llm/names.txt	Does this name exist? (cheapest check)
.llm/resolver.json	Where does this name, alias or URL point?
.llm/entities.jsonl	Signature, parameters, return type.
.llm/namespaces.json	How large is each namespace?
.llm/keywords.json	Which pages mention this term?
.llm/chunks.jsonl	How do I chunk this for retrieval?
.llm/migration.json	What was this called in v5?
.llm/schema.json	What is the exact shape of these records?
.llm/invariants.json	What guarantees does the tree hold?
.graph.json	How are entities connected?

Verify it yourself

// no dependencies beyond the Python standard library
// python3 scripts/validate.py
// python3 scripts/validate.py --json

The validator checks tree depth, kebab-case names, fence languages, fence balance, front-matter keys, _index.md coverage, internal links and SHA-256 integrity.

Provenance and limits

• Built from an offline dump of the official documentation and language reference manual.
• Source language is English and is preserved verbatim. Signatures and reference examples are copied, never paraphrased.
• Guide code blocks were rebuilt from a scraped dump; .llm/docs.jsonl marks them code_fidelity: reconstructed. Reference examples are verbatim.
• Upstream links with no local target stay absolute and are listed in .llm/unresolved.json.
• See NOTICE.md for attribution.

---

Navigation: Master index · Reference · Guide · Migration · Link map · Validation

Tooling

Command	What it does
python3 scripts/find.py ta.sma	Print the signature, parameters, return type and page path
python3 scripts/find.py "moving average"	Concept search over the keyword index
python3 scripts/pack.py --list	List tasks, namespaces and window presets
python3 scripts/pack.py --namespace array	Build one context file for a namespace
python3 scripts/pack.py --task write_a_strategy --out pack.txt	Build a task pack for a model
python3 scripts/validate.py	Prove every structural claim in this repository

New here? Read QUICKSTART.md. Continuous integration runs the validator and smoke-tests both lookup tools on every push.

Contributing

Defect reports are the most useful contribution here, and the issue template asks for exactly what is needed to reproduce a problem. Most of this tree is generated, so read CONTRIBUTING.md before editing files by hand.

python3 scripts/validate.py     # must pass before any pull request

Maintainer and contact

Use the issue tracker for anything about this repository: defects, missing content, or questions about the machine layer. Issues are public and searchable, so they help the next reader as well. For anything that does not fit an issue, use the contact channels on my GitHub profile.

If you represent TradingView and want a change here, open an issue and I will act on it.

License and attribution

The software in scripts/ and .github/, and the repository structure, identifier scheme, JSON schema, knowledge graph and machine layer under .llm/, are released under the MIT License. See LICENSE.

The documentation prose and Pine Script samples under guide/, reference/, migration/, examples/, errors/ and faq/, and their concatenations under dist/, are derived from the official TradingView Pine Script documentation and remain the property of their respective owners. This repository is an unofficial, derived snapshot. It is not affiliated with or endorsed by TradingView. Pine Script and TradingView are trademarks of their respective owners. See NOTICE.md for full attribution, provenance and limits.

Audit trail

• VALIDATION.md publishes the current validator results.
• .llm/link-audit.json records the link and anchor repair history, including the root cause of every defect found before release.