fux-index

Bootstrap, diagnose, and rebuild a Fux search index committed to git.

1|Updated Jun 1, 2026
One-click install
npx skills add https://github.com/arpitarya/fux --skill fux-index-arpitarya
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: fux-index
Source: https://github.com/arpitarya/fux/tree/main/.agents/skills/fux-index
Command: npx skills add https://github.com/arpitarya/fux --skill fux-index-arpitarya

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Setting up and maintaining a Fux search index involves several verbs (setup, doctor, ingest, build) with strict preconditions, lock semantics, and commit rules; this Skill guides an agent through each job so the committed index in .fux/index/ stays valid, current, and byte-identical across clones. ## Core Features & Use Cases - Scaffold and configure: Run fux setup to write fux.toml, source lists, PII rules, and agent files, with controls like --no-agents and [agents] install to limit what is written. - Diagnose before acting: Parse fux doctor --json output, branch on the ok flag and per-row level, and map every failing check row to its documented fix without parsing free-text details. - Build and rebuild the index: Run fux ingest for delta or --full rebuilds, handle the write lock and stale locks, and use fux build to regenerate the derived accelerator that explain, graph, and path require. - Use Case: A teammate says CI fails because the index is stale. You add a workflow step that runs fux doctor, then fux ingest --no-accelerator --no-progress, and gates on git status --porcelain over .fux/index to prove the committed index matches the sources. ## Quick Start Ask the agent to run fux doctor with JSON output in this repository and fix whatever error-level rows it reports.

Frequently Asked Questions about fux-index

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I set up a Fux index in my repository?▼

Run `fux setup` once to scaffold fux.toml, source lists, PII rules, and agent files, then list sources and run `fux ingest` to write the committed index under .fux/index/. Setup is write-if-missing, so existing files are never overwritten.

How do I rebuild a stale Fux index?▼

Run `fux ingest` for a delta rebuild that carries unchanged documents forward by content hash. Use `fux ingest --full` only after a fux upgrade, for the full term-collision check, or to migrate an index written by another fux version.

What should I commit from the .fux directory to git?▼

Commit fux.toml, .fux/index/*.jsonl, sources, formats.toml, .fuxignore, pii.toml, refusals.toml, tune.toml, output.toml, fetchers, decoders, and enrich. Never commit .fux/runtime/ or .fux/acquired/, and never ignore .fux/ wholesale or the index silently leaves git.

Why does fux doctor report warnings but exit with code 0?▼

A warning row has level "warn" and does not affect the exit code; only error-level rows set ok to false and exit 1. Branch on the ok field and each row's level, never on the free-text detail message.

Why does fux refuse to run with a pii.toml missing error?▼

Every verb except setup, doctor, tune, and output stops when .fux/pii.toml is missing because redaction rules are mandatory. Tell the human and run `fux setup` only when asked, since it also writes other committed files and agent files.

How do I verify the Fux index is current in CI?▼

Check out with full history (fetch-depth: 0), run `fux doctor`, then `fux ingest --no-accelerator --no-progress`, and gate on `git status --porcelain` over .fux/index, .fuxignore, and enrich. Do not gate on `fux ingest --check`, whose exit code is always 0.