Obsidian Context Loader

This skill provides tools to extract and load personal context from an Obsidian vault into AI conversations. Use Obsidian's frontmatter tagging system to curate which knowledge gets loaded into your LLM context.

What This Skill Does

Loads markdown notes from an Obsidian vault based on:

Returns chronologically ordered, concatenated content for comprehensive context loading.

Configuration

Before using this skill, create a `config.yaml` file in the same directory as the MCP server:

```yaml

vault_path: "~/Documents/ObsidianVault"

default_context:

properties:

context: "personal"

tags: []

```

Available Tools

1. fetch_context(context_type, chunk_index=0, max_chars=95000)

Load context based on a predefined context type from config.yaml.

**Parameters:**

**Example:**

```python

fetch_context("personal") # Loads all notes with context: personal property

```

2. fetch_matching_files(properties={}, tags=[], match_all_tags=False, chunk_index=0, max_chars=95000)

Flexible runtime filtering by frontmatter properties and tags.

**Parameters:**

**Examples:**

```python

Files with type: project AND containing "active" tag

fetch_matching_files(properties={"type": "project"}, tags=["active"])

Files containing both "research" AND "ai" tags

fetch_matching_files(tags=["research", "ai"], match_all_tags=True)

```

3. fetch_frontmatter_index(properties={}, tags=[], match_all_tags=False)

Browse file metadata without loading full content - ideal for discovery.

**Parameters:**

**Returns:** Formatted table of file metadata (titles, paths, tags, context types)

**Example:**

```python

Browse AI-related notes without loading full content

fetch_frontmatter_index(tags=["ai"])

```

4. fetch_specific_file(file_path)

Retrieve complete content of a specific file by path.

**Parameters:**

**Example:**

```python

After browsing index, load specific file

fetch_specific_file("research/ai-governance-framework.md")

```

5. search_vault_content(search_pattern, case_sensitive=False, regex=False, context_chars=100)

Content-based search through all markdown files (hardcore mode).

**Parameters:**

**Returns:** Frontmatter index of matching files with match context

**Examples:**

```python

Plain text search

search_vault_content("machine learning algorithms")

Regex search

search_vault_content("neural.*network", regex=True)

```

Usage Workflow

Progressive Discovery Pattern (Recommended)

**Phase 1: Browse Metadata**

```python

Get overview of available notes without loading full content

fetch_frontmatter_index(tags=["project"])

```

**Phase 2: Selective Loading**

```python

Load specific files based on metadata review

fetch_specific_file("projects/website-redesign.md")

fetch_specific_file("projects/api-integration.md")

```

Content-Based Discovery

When you don't know the frontmatter structure:

```python

Find notes containing specific content

search_vault_content("API authentication")

Then load relevant files

fetch_specific_file("path/from/search/results.md")

```

Bulk Context Loading

Load all notes matching criteria:

```python

Load all personal finance notes

fetch_matching_files(properties={"context": "personal"}, tags=["finance"])

```

Output Format

Concatenated files use clear delimiters:

```

================================================================================

/absolute/path/to/file.md

================================================================================

---

context: personal

tags: [finance, planning]

---

[full file content]

================================================================================

/absolute/path/to/next/file.md

================================================================================

[full file content]

```

Filtering Logic

**Properties:** AND logic - all specified properties must match exactly (case-sensitive)

**Tags:**

**Combined:** Properties AND tags criteria must both be satisfied

**Sorting:** Files sorted by modification time (oldest first)

Important Notes

Example Frontmatter

```yaml

---

context: personal

type: project

tags: [active, ai, research]

title: AI Research Framework

---

```

Security