#Discovery and indexing
Claudear can auto-discover repositories from your GitHub/GitLab organisations and index them for fast issue routing. Set known_orgs and auto_discover_paths in your core config, then run:
# Discover repos from configured orgs
claudear repos discover
# Index all discovered repos
claudear repos index
# Sync repos (fetch + re-index)
claudear repos sync#Code indexing
Code indexing uses tree-sitter to parse source files and generate embeddings for semantic search. This powers Claudear's ability to route issues to the correct repo and find relevant code when generating fixes.
[code_index]
# Enable tree-sitter code indexing (default: true)
enabled = true
# Maximum file size to index in KB (default: 1024)
max_file_size_kb = 1024
# Embedding batch size (default: 32)
batch_size = 32#Self-evaluation
Before submitting a PR, Claudear can run before/after comparisons on tests, linting, static analysis, and coverage. If the fix introduces regressions, the attempt is flagged.
[evaluation]
# Enable evaluation (default: false, opt-in)
enabled = false
# Run test before/after comparison (default: true)
test_delta = true
# Run lint before/after comparison (default: true)
lint_delta = true
# Run static analysis before/after (default: true)
static_analysis_delta = true
# Run coverage before/after (default: false, slowest)
coverage_delta = false
# Timeout per tool in seconds (default: 300)
tool_timeout_secs = 300
# Total timeout in seconds (default: 900)
total_timeout_secs = 900
# Post results as PR comment (default: true)
post_pr_comment = true
# Fail attempt on regression (default: false)
fail_on_regression = false#Dashboard config
Cost estimation and display settings for the web dashboard:
[dashboard]
# Monthly Claude Max plan cost for per-fix estimates (default: 0.0)
max_plan_monthly_cost = 0.0
# Hourly engineer rate for cost-savings calculation (default: 75.0)
hourly_engineer_rate = 75.0#Dependency graphs
Claudear auto-detects inter-repo dependencies by scanning package manifests (package.json, composer.json) for packages belonging to your known_orgs. The graph is built on first run and refreshed automatically in the background, so new dependencies are picked up without manual intervention.
These relationships power cascading fixes and issue routing. For dependencies that can't be inferred from manifests (e.g. git submodules or cross-org repos), use claudear repos link to declare them manually.
# View the auto-detected dependency graph
claudear repos graph
# Search across indexed repos
claudear repos search "authentication middleware"
# Show repo stats
claudear repos stats#CLI commands
Full repo management commands:
claudear repos discover # Discover repos from orgs
claudear repos index # Index all repos
claudear repos search # Semantic search across repos
claudear repos graph # Show auto-detected dependency graph
claudear repos link # Manually declare a dependency (override)
claudear repos stats # Show indexing statistics
claudear repos sync # Fetch and re-index