Code health, architecture & status

Lead with the one composite signal, then drill only into the weak dimensions and the specific scans the user asked for — don't run every tool by reflex.

Quality scorecard

1. Composite signal → tracedecay_health (details: true, optional path): the 0–10000 score plus the 5-dimension breakdown (acyclicity, depth, equality, redundancy, modularity) and the coverage_discipline penalty. The weak dimensions choose the drill-downs.
2. Inequality / god files → tracedecay_gini (metric: complexity|lines|fan_in|fan_out|members, scope, path?).
3. Complexity & size offenders: tracedecay_complexity, tracedecay_largest, tracedecay_god_class, tracedecay_hotspots.
4. Structure drill-downs (match the weak dimension): acyclicity → tracedecay_circular + tracedecay_recursion; modularity → tracedecay_dsm (format: stats|clusters|matrix) + tracedecay_coupling; depth → tracedecay_dependency_depth + tracedecay_inheritance_depth; relationships → tracedecay_rank (edge_kind required); kind mix → tracedecay_distribution.
5. Duplication → tracedecay_redundancy; near-duplicate names → tracedecay_similar. Doc gaps → tracedecay_doc_coverage. Panic sites → tracedecay_unsafe_patterns. Risk-weighted test gaps → tracedecay_test_risk. Changed-files-only pass → tracedecay_simplify_scan (files).

Architecture map

1. Shape & size: tracedecay_status (node/edge/file counts), tracedecay_files + tracedecay_distribution (what lives where).
2. Public surface: tracedecay_module_api per top-level directory.
3. Dependency structure: tracedecay_dsm (clusters and layering violations), tracedecay_coupling (fan_in/fan_out hubs), tracedecay_circular (cycles), tracedecay_dependency_depth (fragile long chains).

Session health delta

1. Before the first edit → tracedecay_session_start (no args): snapshots current health as the baseline (.tracedecay/session_baseline.json).
2. After the work → tracedecay_session_end: the per-dimension diff — what improved, what degraded — and clears the baseline. A dropped dimension names the follow-up (redundancy fell → tracedecay_redundancy; acyclicity fell → tracedecay_circular).
3. Bracket only work where a before/after delta is wanted; a second session_start silently overwrites the baseline.

Project & index status

1. Active project → tracedecay_active_project (no args): resolved project root, scope prefix, branch identity, and the resolved active project store backing this session. Use this before describing where data lives.
2. Storage status → tracedecay_storage_status (no args): resolved active project store health, graph DB path, writability, branch-fallback warnings — instead of probing .tracedecay or direct SQLite checks.
3. Project registry / cross-project context → tracedecay_project_list / tracedecay_project_search / tracedecay_project_context when the user asks about another project, sibling workspace, cross-repo context, or cross-project context gathering. Confirm the target store first, then pass project_id, project_path, or project_selector to tracedecay_context, tracedecay_search, or tracedecay_message_search instead of scanning parent directories.
4. Index status → tracedecay_status: node/edge/file counts, DB size, active branch + fallback warning, tokens saved.
5. Config lookups → tracedecay_config (key required, plus path or glob): query TOML/JSON by dotted key — works even before tracedecay init.
6. Outstanding work → tracedecay_todos (kinds?, path?, limit?).
7. Server triage → tracedecay_runtime (PID, memory, CPU%, DB sizes) when TraceDecay seems to hog CPU or RAM. Visual → tracedecay_dashboard (action: start|stop): hand the URL to the user.

Guardrails

• Discovery/analysis tools are read-only and parallel-safe. tracedecay_session_start/session_end write/remove the baseline file and tracedecay_dashboard starts/stops a local server — use them only when relevant and respect Cursor approval/run-mode.
• tracedecay_redundancy is computed lazily and cached; the first call on a fresh index can be slow — keep path/max_pairs tight.
• For large audits, use scoped read-only subagents by path, weak health dimension, or top-level directory; keep session_start/session_end in the parent agent.
• This skill reports and prioritizes; it does not edit. Hand fixes to tracedecay:editing-safely / tracedecay:reviewing-changes, verification to tracedecay:assessing-impact. Memory recall belongs to tracedecay:project-memory; past-session recall to tracedecay:managing-session-context.

If tools are deferred or MCP fails

• Deferred (names listed without schemas): load once with ToolSearch — select:tracedecay_health,tracedecay_gini,tracedecay_dsm,tracedecay_status,tracedecay_active_project,tracedecay_project_list,tracedecay_project_search,tracedecay_project_context (one batched call, add others needed) — then call normally.
• MCP error/timeout/disconnect: same tool, same args, via shell: tracedecay tool <name> (see tracedecay:using-the-cli). Never query .tracedecay databases directly; never abandon the graph over transport.

Deliverable

Do not end this workflow without: the composite score + weak dimensions with ranked worst offenders and a prioritized fix list; the layered module map with dependency hotspots/violations; the per-dimension session delta; or the status numbers, config values (with file + line), marker list, or runtime snapshot the user asked for. Pairs with the docs-canvas plugin if installed. Report any tracedecay_metrics: line to the user.