get_change_intelligence

Tells you which symbols are dangerous to change by blending two signals the index already has: cyclomatic complexity (from the parser) and churn (commit count from git). High score = hotspot.

When to use

Run this on any symbol before a non-trivial edit. Pair it with get_blast_radius to understand how far a change spreads, and with get_why_context to learn why the area is risky.

Parameters

Name	Type	Required	Description
symbolId	string	yes	Fully-qualified symbol id (<file>::<name>::<kind>)

Example

{
  "symbolId": "packages/cli/src/adapters/storage/sqlite.ts::SqliteStorageAdapter::class"
}

Response

{
  "symbolId": "packages/cli/src/adapters/storage/sqlite.ts::SqliteStorageAdapter::class",
  "complexity": 0.78,
  "churn": 0.92,
  "composite": 0.85,
  "band": "high"
}

Symbol missing from the index:

{ "found": false, "hint": "Symbol not found. Run \"ctxo index\" to build the codebase index." }

Interpreting the score

Field	Range	Meaning
complexity	0.0-1.0	Normalized cyclomatic complexity (1 -> 0.0, 10+ -> 1.0)
churn	0.0-1.0	Normalized commit count vs the hottest file in the repo
composite	0.0-1.0	Weighted blend
band	enum	low (< 0.3), medium (0.3-0.7), high (>= 0.7)

When to use

• Refactor planning -- rank candidates by composite score; start with high-band symbols only if you have a test net.
• Code review triage -- a diff that touches multiple high-band symbols deserves a closer look.
• Hotspot discovery -- combined with get_symbol_importance, you get the "critical and volatile" shortlist.

Pitfalls

Class-level aggregation

For a class or file symbol, the handler aggregates complexity as the max across its methods. A class can score high even if most methods are trivial. Drill down with search_symbols to inspect the offending method directly.

• Drifted index -- if the JSON index is missing for the file, complexity falls back to 0 and a warning is surfaced alongside the response. Do not mistake this for a real "low" verdict. Run ctxo index to resync.
• Git not found -- churn will be 0. The score stops being meaningful.
• No unit here is "commits per week" -- values are normalized against the repo's busiest file.

Related tools

• get_blast_radius -- scope of a change
• get_why_context -- why the area is risky
• get_symbol_importance -- structural centrality
• get_pr_impact -- change intelligence rolled up across a PR