docs(observability): inventory component coverage

[tucktuck101]

Summary

Adds an 80-row canonical component inventory with stable IDs, assessment owners, pinned evidence, and explicit product boundaries. It now distinguishes pending assessment work from evidence-backed bounded unknowns, preserving PRD #289's epistemic states while making stage-1 progress truthful. Current-state navigation identifies the matrix as the coverage and assessment-progress record while preserving mobile and infrastructure exclusions.

Related issue

Closes #462

Issue type

Task

---

Agent provenance

Field	Value
Harness / provider	Codex
Model	gpt-5.6-sol
Session reference	N/A - the harness exposes no shareable run id
Initiating human	@tucktuck101

Objective

Create the canonical non-mobile Buzz component coverage matrix and make it the linked stage-1 coverage and assessment-progress record.

Impacted components

launchpad/docs/Observability/README.md
launchpad/docs/Observability/current-state/coverage.md
launchpad/docs/Observability/current-state/overview.md
launchpad/docs/Observability/current-state/relay.md
launchpad/docs/Observability/current-state/desktop.md
launchpad/docs/Observability/current-state/web.md

Approach and rejected alternatives

The matrix gives each product boundary a stable ID and one assessment owner, then uses separate reconciliation registers for enumerable repository surfaces. Source registration proves coverage and ownership, not component-level runtime evidence, so incomplete work is Pending assessment; only a completed assessment may produce Assessed, a specifically evidenced Bounded unknown, or Excluded.

Rejected using Bounded unknown as a synonym for work not yet performed because PRD #289 explicitly distinguishes an examined-but-unresolved property from an unexamined component. Rejected a one-row-per-crate inventory because relay and desktop packages host many independently diagnosable boundaries; shared libraries remain assigned through consuming runtimes unless they own a distinct diagnostic or export seam.

Verification

Command run:

set -euo pipefail
awk -F'|' '
/^\| [A-Z][0-9][0-9] / {
  id=$2; owner=$7; evidence=$8; state=$9; reason=$10;
  gsub(/^ +| +$/, "", id); gsub(/^ +| +$/, "", owner);
  gsub(/^ +| +$/, "", evidence); gsub(/^ +| +$/, "", state);
  gsub(/^ +| +$/, "", reason); rows++;
  if (NF != 11) { print "bad-column-count " id " NF=" NF; errors++ }
  if (seen[id]++) { print "duplicate-id " id; errors++ }
  if (state != "Assessed" && state != "Bounded unknown" && state != "Excluded" && state != "Pending assessment") { print "invalid-state " id " state=" state; errors++ }
  if (state == "Pending assessment" && owner !~ /issues\/(46[3-9]|47[0-6])/) { print "pending-without-active-owner " id " owner=" owner; errors++ }
  if (state == "Bounded unknown" && reason !~ /(unknown|could not|not established)/) { print "unbounded-unknown " id; errors++ }
  if (evidence !~ /(26920e5c|678008ea)/) { print "missing-revision " id; errors++ }
  if (reason == "") { print "missing-reason " id; errors++ }
  states[state]++;
}
END {
  printf "rows=%d Assessed=%d Pending=%d Bounded_unknown=%d Excluded=%d errors=%d\n", rows, states["Assessed"], states["Pending assessment"], states["Bounded unknown"], states["Excluded"], errors;
  exit(errors != 0)
}' launchpad/docs/Observability/current-state/coverage.md

. ./bin/activate-hermit >/dev/null 2>&1
inventory_actual=$( { cargo metadata --no-deps --format-version 1 | jq -r '.packages[].name'; cargo metadata --no-deps --format-version 1 --manifest-path desktop/src-tauri/Cargo.toml | jq -r '.packages[].name'; } | sort -u )
inventory_documented=$(awk '/^\| Package or target / { table=1; next } table && /^\|---/ { next } table && /^\| `/ { print; next } table && !/^\|/ { exit }' launchpad/docs/Observability/current-state/coverage.md | sed -E 's/^\| `([^`]+)`.*/\1/' | sort -u)
if [ "$inventory_actual" != "$inventory_documented" ]; then
  diff -u <(printf '%s\n' "$inventory_actual") <(printf '%s\n' "$inventory_documented")
  exit 1
fi
printf 'cargo-package-register: ok (%s packages)\n' "$(printf '%s\n' "$inventory_actual" | wc -l | tr -d ' ')"

for doc_target in \
  launchpad/docs/Observability/current-state/coverage.md \
  launchpad/docs/Observability/current-state/overview.md \
  launchpad/docs/Observability/current-state/relay.md \
  launchpad/docs/Observability/current-state/desktop.md \
  launchpad/docs/Observability/current-state/web.md; do
  test -f "$doc_target" || { printf 'missing-link-target: %s\n' "$doc_target"; exit 1; }
done
defined_ids=$(awk -F'|' '/^\| [A-Z][0-9][0-9] / { id=$2; gsub(/^ +| +$/, "", id); print id }' launchpad/docs/Observability/current-state/coverage.md | sort -u)
referenced_ids=$(rg -o '\b[A-Z][0-9]{2}\b' launchpad/docs/Observability/current-state/coverage.md | sort -u)
unknown_ids=$(comm -13 <(printf '%s\n' "$defined_ids") <(printf '%s\n' "$referenced_ids"))
if [ -n "$unknown_ids" ]; then
  printf 'undefined-component-ids:\n%s\n' "$unknown_ids"
  exit 1
fi
printf 'doc-link-targets: ok (5 files)\ncomponent-id-references: ok (%s defined IDs)\n' "$(printf '%s\n' "$defined_ids" | wc -l | tr -d ' ')"
git diff --check

Raw output:

rows=80 Assessed=5 Pending=70 Bounded_unknown=0 Excluded=5 errors=0
cargo-package-register: ok (33 packages)
doc-link-targets: ok (5 files)
component-id-references: ok (80 defined IDs)

• Tests or checks were run and the raw output is pasted above
• The diff is confined to the scope of the linked issue
• No secrets, keys, tokens or hostnames were added to tracked files

Not verified

No relay, desktop, browser, ACP, MCP, CLI, worker, adapter, or subprocess workflow was run; issue #462 inventories and assigns those components while issues #463-#476 own their current-state assessments. No deployed collection or infrastructure system was inspected, and no mobile internals were assessed. The admin-web bundle was verified from source and release packaging but not exercised in a browser.

Security implications

No runtime behavior or exposure changes. The documentation records security-sensitive diagnostic boundaries and keeps dependency internals, mobile, and post-export infrastructure outside the product-instrumentation scope; inspection found no credential values or private hostnames added by this diff.

Escalations

The original #462 definition of done used Bounded unknown for not-yet-performed assessments, conflicting with PRD #289. The issue contract and this matrix are corrected to use transitional Pending assessment; Gate 2 must reduce that count to zero from evidence produced by #463-#476. Separately, #457-#460 were merged before their recorded predecessors completed; that orchestration deviation remains for the human Gate 1 review and is not silently resolved in this inventory PR.