Skip to content
Skills Library
AI & agents · Tool integration

cli-vstash

/cli-vstash

Local document memory with semantic search for AI-assisted workflows. Use when managing project documentation, codebases, or research papers that need persistent memory across sessions. Triggers on: vstash add/search/ask commands, document ingestion, semantic search, RAG pipelines, local knowledge bases, or configuring vstash for personal projects.

Repository
prompts ↗
Licence
MIT
Last repository change
02.09.2026

Assessment and security

100/ 100Class: good
Format25 / 25
Frontmatter with name and description, valid name, description up to 1,024 characters, name matching the folder.
Discoverability25 / 25
Description explains when to use the skill, has a reasonable length and uses third person.
Content35 / 35
At least 150 words and at most 500 lines, headings, step lists, examples or code.
Hygiene15 / 15
No placeholders such as TODO, no Windows paths, with a licence.
Riskno signals detected in the scan
Files in this package
2
Scripts
0
HTML comments
0
URLs in SKILL.md
2
Deprecation

No signals detected. Repository changed on 02.09.2026, without references to retired models or APIs.

Licence

MIT, open: use and modify while retaining author attribution.

What it does and when to use it

Local document memory with semantic search for AI-assisted workflows. Use when managing project documentation, codebases, or research papers that need persistent memory across sessions. Triggers on: vstash add/search/ask commands, document ingestion, semantic search, RAG pipelines, local knowledge bases, or configuring vstash for personal projects.

Original author description

Local document memory with semantic search for AI-assisted workflows. Use when managing project documentation, codebases, or research papers that need persistent memory across sessions. Triggers on: vstash add/search/ask commands, document ingestion, semantic search, RAG pipelines, local knowledge bases, or configuring vstash for personal projects.

How it works

  1. Core Commands
  2. Ingestion Pipeline
  3. Search Pipeline
  4. Configuration
  5. Metadata Filtering
  6. Privacy
  7. Reference
Original SKILL.md headings · 123 lines, 445 words
SKILL.md preview
---
name: cli-vstash
description: 'Local document memory with semantic search for AI-assisted workflows. Use when managing project documentation, codebases, or research papers that need persistent memory across sessions. Triggers on: vstash add/search/ask commands, document ingestion, semantic search, RAG pipelines, local knowledge bases, or configuring vstash for personal projects.'
---

# vstash CLI

Local document memory with instant semantic search. Drop any file, ask anything.

## Core Commands

```bash
# Ingest documents (PDF, DOCX, PPTX, XLSX, Markdown, code, URLs)
vstash add paper.pdf notes.md https://example.com/article
vstash add ./docs --collection research --project ml-survey

# Semantic search (free, no LLM needed)
vstash search "what's the main argument about X?"

# Ask with LLM (requires inference backend)
vstash ask "summarize the key findings"

# Interactive chat session
vstash chat

# Document management
vstash list                          # Show all ingested documents
vstash stats                         # DB statistics (docs, chunks, size)
vstash forget paper.pdf              # Remove document from memory
vstash reindex                       # Switch embedding model without re-ingesting

# Auto-ingest on file changes
vstash watch ./docs

# Export for training data curation
vstash export --project ml-survey --format jsonl

# Show current configuration
vstash config
```

## Ingestion Pipeline

```text
file/URL → parse → chunk → embed → store vectors → index text
```

**Parsing:**

- Non-code: markitdown (preserves structure)
- Code (`code_aware=true`): raw UTF-8 to preserve syntax

**Chunking strategies:**

| Mode | Files | Strategy |
|------|-------|----------|
| Semantic | MD, PDF, DOCX | Headers → paragraphs → fixed-window → merge small |
| Code-aware | Python, JS/TS, Go, Rust, Java | Split at top-level `def`/`class`/`func`/`fn` |

## Search Pipeline

```text
query → embed → vector search (top-k×10) → keyword search (top-k×10) → RRF fusion → memory scoring → dedup → relevance signal → top-k
```

**Reciprocal Rank Fusion (RRF):** Merges vector + keyword rankings without comparable scores.

**Relevance signal:** Distance-based confidence tiers:

- ≤ 0.95: high (full confidence)
- 0.95–0.98: medium (uncertain)
- > 0.98: low (results may not match)

**Context expansion:** ±1 adjacent chunks included for LLM answers (2.64× more context).

## Configuration

Create `vstash.toml` in current directory or `~/.vstash/vstash.toml`:

```toml
[inference]
backend = "ollama"  # or "cerebras", "openai"

[ollama]
host = "http://localhost:11434"
model = "llama3.2"

[embeddings]
model = "BAAI/bge-small-en-v1.5"  # multilingual: "BAAI/bge-m3"

[chunking]
size = 1024
overlap = 128
code_aware = true

[scoring]
enabled = true  # frequency + temporal decay re-ranking
```

Run `vstash config` to see active settings.

## Metadata Filtering

```bash
vstash add notes.md --collection research --project ml-survey --tags "attention,transformers"
vstash list --project ml-survey
vstash ask "what architectures were compared?" --project ml-survey
```

Documents with YAML frontmatter are parsed automatically.

## Privacy

| Component | Local? |
|-----------|--------|
| Embeddings (FastEmbed ONNX) | Yes |
| Vector store (sqlite-vec) | Yes |
| Semantic search | Yes |
| Inference (Ollama) | Yes |
| Inference (Cerebras/OpenAI) | No — chunks sent to API |

For full privacy, use `backend = "ollama"` or use `vstash search` instead of `vstash ask`.

## Reference

See `references/cli-reference.md` for full option details.
Files in this package (2)
  • SKILL.md 4 KB
  • references/cli-reference.md 3 KB

Versions

The database has no other skills named cli-vstash.

Database snapshot: 13.11.2026 · package version 179660a26b6a (SHA-256 digest of the entire package)

Downloads in the last 30 days: 0 · ranking

Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.

Similar skills

  • 95mcp-vstash
    MCP server integration for vstash document memory. Use when configuring Claude Desktop or other MCP-compatible AI assistants with persistent document memory, setting up vstash MCP tools for semantic search and Q&A, or integrating vstash with AI assistant workflows via Model Context Protocol.
  • 100langchain4j-rag-implementation-patterns
    Provides Retrieval-Augmented Generation (RAG) implementation patterns with LangChain4j for Java. Generates document ingestion pipelines, embedding stores, vector search, and semantic search capabilities. Use when building chat-with-documents systems, document Q&A over PDFs or…
  • 100rag-implementation
    Build Retrieval-Augmented Generation (RAG) systems for LLM applications with vector databases and semantic search. Use when implementing knowledge-grounded AI, building document Q&A systems, or integrating LLMs with external knowledge bases.
  • 100rag-implementation
    Build Retrieval-Augmented Generation (RAG) systems for LLM applications with vector databases and semantic search. Use when implementing knowledge-grounded AI, building document Q&A systems, or integrating LLMs with external knowledge bases.
  • 100rag-implementation
    Build Retrieval-Augmented Generation (RAG) systems for LLM applications with vector databases and semantic search. Use when implementing knowledge-grounded AI, building document Q&A systems, or integrating LLMs with external knowledge bases.
  • 100rag-implementation
    Build Retrieval-Augmented Generation (RAG) systems for LLM applications with vector databases and semantic search. Use when implementing knowledge-grounded AI, building document Q&A systems, or integrating LLMs with external knowledge bases.

Feedback and reports

Your message goes to Bartosz Sokoliński. An email address is needed only for a reply. See the privacy policy.