06 — DOT / Tool Count Reliability

Counts (live)

metric	value
dot_tools	309
dot_tools non-retired	309
wf_fs_dot_bin_snapshot (host /opt/incomex/dot/bin)	287 (≈288 files)
wf_fs_script_snapshot (/opt/incomex/scripts)	42
dot_iu_command_catalog	54
v_dot_registry_no_file	41
v_dot_fs_reconciliation	287
dot_tools names matching noise pattern (.bak/backup/retired/_old/pre/deprecat)	3

The core DOT reconciliation gap

• dot_tools = 309, but only ~119 have a script_path (per v_workflow_discovery_source_missing_adapter note for fs_dot_bin). The remaining ~190 dot_tools are DB-resident/non-file tools.
• v_dot_registry_no_file = 41 — 41 registry entries (13%) reference files that do not exist on disk. → stale registry rows or DB-only tools.
• 287 bin files on disk vs ~268 registry entries with files → ~18–19 files unregistered (matches the WPC-DOT-BIN-RECONCILE=18 process candidate).

→ The headline dot_tools=309 is a reliable REGISTRY count, but NOT a reliable count of live executables. As "how many DOT tools exist" it is fine; as "how many runnable scripts" it overstates by the 41 missing-file entries and is structurally disconnected from the 287 bin files. The reconciliation views surface all of this (registry_no_file, fs_reconciliation) — the gap is visible, not hidden.

Noise / duplicates / stale

• Only 3 dot_tools names look like backup/retired/noise → low in-registry noise.
• Backup/retired script artifacts are corralled separately as the WPC-NOISE-QUARANTINE process candidate (3) on the host side, not inside dot_tools.
• No duplicate-name explosion detected in dot_tools.

DOT process-like Type-1 candidates vs helpers/tools/noise

The 309 dot_tools are mostly implementation components (Type-1 building blocks), not standalone processes. The process layer correctly does NOT treat each dot_tool as a process (official AX-PROCESS=0); only 3 DOT_IMPLEMENTATION_ONLY clusters (DOT-BIN-RECONCILE 18, DOT-NRM 14, DOT-HC 2) are surfaced as candidates. Good separation of helper vs process.

dot_iu_command_catalog = 54

Matches census dot_iu_command raw_count 54 (LIVE_DB_FN). Reliable.

Classification

• dot_tools=309 registry count → OK; as live-executable count → COUNT_UNRELIABLE (interpret-with-care).
• 41 registry_no_file → STALE_DATA (surfaced).
• ~18 file_no_registry → CANDIDATE_EXPECTED (already a remediation candidate).

Score: 68/100

Clean registry, low noise, reconciliation gaps fully surfaced. The lowest base-DB score because the headline tool count and the real executor layer (host bin + registry_no_file + script_path coverage) are only partially reconciled, and the authoritative schedule lives off-DB (see doc 08).