Graph Export Skill

Seamless access to wicked-search graph data for other plugins.

Quick Start

python

from graph_client import GraphClient

# Initialize with workspace path (auto-connects to wicked-cache)
client = GraphClient("/path/to/workspace")

# Check freshness before querying
if not client.is_fresh(max_age_seconds=3600):
    print("Graph data may be stale - consider re-indexing")

# Query symbol dependencies
result = client.get_symbol_dependencies()
for symbol in result.symbols:
    print(f"{symbol.name}: {len(symbol.dependencies)} deps, {len(symbol.dependents)} dependents")

# Lookup definition by name
location = client.lookup_definition(name="UserService")
if location:
    print(f"Found at {location.file}:{location.line_start}")

# Get call chain (blast radius)
chain = client.get_call_chain("src/auth.py::login", max_depth=3)

Available Query Types

Method	Returns	Use Case
`get_symbol_dependencies()`	All symbols with deps/dependents	Dependency analysis
`get_file_references(files)`	File-level symbol summaries	File impact analysis
`lookup_definition(name)`	Symbol location	Jump-to-definition
`get_call_chain(id, depth)`	Upstream/downstream chains	Blast radius

Filter Support

All queries support optional filters (workspace-scoped by default):

python

# Filter by paths
deps = client.get_symbol_dependencies(filter={
    "paths": ["src/auth/"],
    "exclude_paths": ["src/auth/tests/"],
    "node_types": ["FUNCTION", "METHOD"],
    "domain": "code"
})

# Filter file references
refs = client.get_file_references(files=["src/api.py", "src/models.py"])

Freshness & Versioning

python

# Check if data is fresh (default: 1 hour)
is_ok = client.is_fresh(max_age_seconds=3600)

# Get detailed freshness info
meta = client.get_freshness()
print(f"Indexed at: {meta.indexed_at}")
print(f"Files: {meta.file_count}, Nodes: {meta.node_count}")

Error Handling

python

from graph_client import GraphClient, CacheStaleError, VersionMismatchError

try:
    result = client.get_symbol_dependencies()
except CacheStaleError:
    print("Cache miss - run /wicked-search:index first")
except VersionMismatchError as e:
    print(f"Schema changed: {e}")

Integration Patterns

For Plugin Authors

Add this to your plugin's requirements:

code

wicked-search  # Provides graph-export skill
wicked-cache   # Cache infrastructure

In your skill/agent, use the client:

python

import sys
sys.path.insert(0, "${WICKED_SEARCH_SCRIPTS}")
from graph_client import GraphClient

Triggering Cache Export

To populate the cache, run indexing with --export-cache:

bash

cd ${CLAUDE_PLUGIN_ROOT}/scripts && uv run python unified_search.py index /path/to/project --export-graph --export-cache

This will:

•Build the JSONL index (fast, parallel)
•Build the Symbol Graph (JSP/HTML support)
•Export to wicked-cache for cross-plugin access

For manual export (advanced):

python

from graph_export import GraphExporter
from symbol_graph import SymbolGraph

exporter = GraphExporter(cache)
result = exporter.export_all(graph, workspace_path)
print(f"Exported {len(result.keys_written)} cache entries")

References

•Cache Schema - Full schema specification
•Usage Examples - Real-world patterns