zift

AI Community

Repo stars 39
Author repo awesome-omni-skill

Interpretation is structured for decision-making; original keeps the upstream SKILL.md unchanged.

Decide Fit First

Core job: Fast, semantic, and hybrid code search tool. Use when you need to find specific code patterns, understand architectural flows, o…
Best fit: Use it when the task has reusable inputs, steps, and validation criteria rather than a one-off answer.
Avoid forcing it: If the source lacks commands, platform support, or external-service evidence, keep those fields unknown instead of guessing.

Design Intent

Structure: The skill is organized around “Core Search Modes”, “1. Semantic / Hybrid (Default)”, “2. Exact Match (-e, --exact)”, “3. Regex Search (-r, --regex)”, showing how the author expects the agent to judge fit, collect context, and produce verifiable output.
Trigger evidence: Prioritize the author’s wording around when to use it, what context to collect, and what output shape to produce.
Evidence boundary: Author text states facts, repository files prove commands and paths, and Fluxly only adds fit, limits, and usage judgment.

How To Use It

Inputs: Provide target material, scope, expected result, forbidden changes, and validation method.
Invocation: Name zift directly; if the source includes slash commands, start with the command and then add task context.
Validation: Start small and check whether the result follows “Core Search Modes / 1. Semantic / Hybrid (Default) / 2. Exact Match (-e, --exact)” before expanding.

Boundaries And Review

Dependencies: It usually needs no extra API key, so start with a small validation task.
Permissions: Declared permissions include read; ask the agent to state file, command, and rollback boundaries before acting.
Quality bar: A useful result names the deliverable, evidence, and next action. Generic prose means the task needs tighter context.

zift Skill

zift is a high-performance, local-first code search tool that combines semantic (vector) and lexical (FTS5) search.

Core Search Modes

1. Semantic / Hybrid (Default)

Uses natural language to find code by "meaning" or "intent".

zift "how do I handle database connections" .

Best for: High-level architectural questions, finding unfamiliar logic.

2. Exact Match (`-e`, `--exact`)

Performs a literal, byte-perfect substring search.

zift -e "SearchResult { file_path" .

Best for: Finding specific variable names, error strings, or boilerplate.

3. Regex Search (`-r`, `--regex`)

Uses Rust-powered regular expressions for pattern matching.

zift -r "pub (async )?fn" .

Best for: Finding all instances of a pattern (e.g., all public functions).

Workflow

Indexing

Before searching, you must index the project. zift uses incremental indexing and will auto-refresh stale files on query, but a full initial index is recommended:

zift add .

Filtering

Use the -l or --local flag to restrict results to the current working directory (useful in monorepos).

zift "auth logic" . --local

Architecture & Performance

Local-first: All embeddings and indices stay on your machine (~/.cache/zift).
GPU Accelerated: Uses Metal/GPU for embedding generation via llama-cpp-2.
Hybrid RRF: Fuses semantic and lexical results using Reciprocal Rank Fusion (k=60) with Power-Law scaling (γ=2.5) for intuitive percentages.
Speed: Exact and Regex searches skip the embedding phase and are near-instant (<50ms).

Troubleshooting

Stale Index: If results seem old, run zift add . again or zift forget . to reset.
Model Issues: zift defaults to nomic-embed-text-v1.5. Ensure the model is downloaded to the cache directory.

Fluxly profile Author and license come from source; runtime, permissions, and network are Fluxly detections or estimates

Fluxly category: AI
Author-declared agents: No explicit declaration found; this is not inferred or tested compatibility
Static check: 88 / 100 · heuristic scan, not runtime safety proof
Author / version / license: @diegosouzapw · no license declared
Fluxly token estimate: Lean
Fluxly setup estimate: Plug-and-play
External API key: No requirement detected
Detected OS requirements: Unspecified
Runtime requirements: Unspecified
Detected file/system behavior: Read-only
Detected network behavior: Local-only
Install commands: None (reference only)

Profile is derived at build time from SKILL.md and install vectors. Subject to drift from author intent.

Heads up: 未限定 allowed-tools，默认拥有全部工具权限。

Output preview zift.preview

The current SKILL.md does not define a fixed output example.

SKILL.md SKILL.md

# zift Skill

`zift` is a high-performance, local-first code search tool that combines semantic (vector) and lexical (FTS5) search.

## Core Search Modes

### 1. Semantic / Hybrid (Default)
Uses natural language to find code by "meaning" or "intent".
```bash
zift "how do I handle database connections" .
```
*Best for: High-level architectural questions, finding unfamiliar logic.*

### 2. Exact Match (`-e`, `--exact`)
Performs a literal, byte-perfect substring search.
```bash
zift -e "SearchResult { file_path" .
```
*Best for: Finding specific variable names, error strings, or boilerplate.*

### 3. Regex Search (`-r`, `--regex`)
Uses Rust-powered regular expressions for pattern matching.
```bash
zift -r "pub (async )?fn" .
```
*Best for: Finding all instances of a pattern (e.g., all public functions).*

## Workflow

### Indexing
Before searching, you must index the project. `zift` uses incremental indexing and will auto-refresh stale files on query, but a full initial index is recommended:
```bash
zift add .
```

### Filtering
Use the `-l` or `--local` flag to restrict results to the current working directory (useful in monorepos).
```bash
zift "auth logic" . --local
```

## Architecture & Performance
- **Local-first**: All embeddings and indices stay on your machine (`~/.cache/zift`).
- **GPU Accelerated**: Uses Metal/GPU for embedding generation via `llama-cpp-2`.
- **Hybrid RRF**: Fuses semantic and lexical results using Reciprocal Rank Fusion (k=60) with Power-Law scaling (γ=2.5) for intuitive percentages.
- **Speed**: Exact and Regex searches skip the embedding phase and are near-instant (<50ms).

## Troubleshooting
- **Stale Index**: If results seem old, run `zift add .` again or `zift forget .` to reset.

…

Evidence boundary and execution chain

Author text anchors workflow facts; Fluxly only indexes current sections, terms, files, and commands.

sections -> Core Search Modes → 1. Semantic / Hybrid (Default) → 2. Exact Match (-e, --exact) → 3. Regex Search (-r, --regex) → Workflow → Indexing
terms    -> Local-first · GPU Accelerated · Hybrid RRF · Speed · Stale Index · Model Issues
files/cmd -> zift · --exact · --regex · --local · ~/.cache/zift · llama-cpp-2 · zift add . · zift forget .
body sha256 -> f9b3d563301f