repl: :cost meta surface (Phase 7 commit #5)

User-facing reporter of the per-session accumulator. Three shapes:

  :cost            one-line summary (calls / tokens / cost)
  :cost detail     per-model + per-category breakdown
  :cost reset      zero the meter; clears warn flags

All read-only against ctx.usage_totals; no broker calls.

R6 — annotation uses the per-slot is_local sticky flag, NOT a fragile
cost==0 heuristic. Summary line classifies:

  cloud only -> "cost=$X.XXXXXX"
  cloud + local mix -> "cost=$X.XXXXXX (cloud only; local: tokens
                       but no cost field)"
  local only -> "cost=$X.XXXXXX (local only; no cost field)"

R7 — :cost detail rows sort by (cost desc, model asc, category asc).
Three-level key for deterministic output across equal-cost rows
(table.sort is unstable; identical costs would otherwise reorder).

R10 — all dollar values use $%.6f formatting. Sub-cent precision is
critical: a Haiku call can cost $0.000028; $%.4f would round it to
$0.0000 — indistinguishable from local $0.

Column width widened to %-26s to fit fully-qualified cloud model
names (e.g. "anthropic/claude-haiku-4.5" = 25 chars).

E2E verified against live cloud + local broker:

  :cost (empty session)          -> "0 calls, $0.000000"
  ...after mixed-mode session...
  :cost                          -> "5 calls, prompt=472 / completion=26
                                     tokens, cost=$0.000377 (cloud only;
                                     local: tokens but no cost field)"
  :cost detail                   -> 4 rows: main cloud $0.000219, probe
                                     cloud $0.000128, delegate cloud
                                     $0.000030, main local $0.000000
                                     (local). Sort by cost desc within
                                     model.
  :cost reset                    -> "cost meter reset"; subsequent
                                     :cost shows zeros.

All 5 categories appeared in the same session: main (twice — cloud
+ local), delegate, probe (x2 from :safety check). Warn-threshold
firing already verified in commit #3 + #4.

HELP gains 3 :cost lines.

Regression: test_safety 87/87, test_router_model 31/31, repl loads.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

repl.lua

@@ -185,6 +185,9 @@ Meta commands:
   :tree off               clear the [project] block
   :diff [<git-args>]      git diff <args> -> inject as [diff ...] exec_output
                           examples: :diff   :diff --cached   :diff main..feature
+  :cost                   summary of session token/cost usage
+  :cost detail            per-model + per-category breakdown
+  :cost reset             zero the cost meter (also clears warn flags)
   :highlight [on|off|status]
                           toggle tree-sitter syntax highlighting on assistant
                           code fences (requires external tree-sitter CLI +
@@ -2024,6 +2027,78 @@ function M.run(config)
     -- Phase 6: :diff meta — `git diff <args>` (B1-clean), appends as
     -- [diff <args>]\n<output> exec_output. Reads cwd at invocation
     -- time (R6: differs from :tree's scan-time cwd capture). Empty
+    -- Phase 7: :cost meta — read-only reporter of ctx.usage_totals.
+    --   :cost            summary line
+    --   :cost detail     per-model + per-category breakdown
+    --   :cost reset      zero the meter (also clears warn flags)
+    -- R7 sort: (cost desc, model asc, category asc) — table.sort is
+    -- unstable, so the 3-level key ensures deterministic output.
+    -- R10: $%.6f for sub-cent precision (cloud costs can be ~3e-05).
+    -- R6: annotation uses the per-slot is_local sticky flag rather
+    -- than a fragile cost==0 heuristic.
+    meta.cost = function(args)
+        local sub = ((args or ""):match("^%s*(%S*)") or ""):lower()
+        if sub == "reset" then
+            ctx:reset_usage()
+            renderer.status("cost meter reset")
+            return
+        end
+        local total_cost  = ctx:total_cost()
+        local total_p, total_c = ctx:total_tokens()
+        local has_local, has_cloud = false, false
+        for _, m in pairs(ctx.usage_totals or {}) do
+            for _, c in pairs(m) do
+                if c.is_local then has_local = true end
+                if c.cost > 0 then has_cloud = true end
+            end
+        end
+        local label
+        if has_local and has_cloud then
+            label = "(cloud only; local: tokens but no cost field)"
+        elseif has_local and not has_cloud then
+            label = "(local only; no cost field)"
+        else
+            label = ""
+        end
+        if sub == "" then
+            local calls = 0
+            for _, m in pairs(ctx.usage_totals or {}) do
+                for _, c in pairs(m) do calls = calls + c.calls end
+            end
+            renderer.status(("session usage: %d calls, prompt=%d / completion=%d tokens, cost=$%.6f %s"):format(
+                calls, total_p, total_c, total_cost, label))
+            return
+        end
+        if sub == "detail" then
+            local rows = {}
+            for model, cats in pairs(ctx.usage_totals or {}) do
+                for category, c in pairs(cats) do
+                    rows[#rows + 1] = {
+                        model = model, category = category,
+                        prompt = c.prompt, completion = c.completion,
+                        calls  = c.calls,  cost = c.cost,
+                        is_local = c.is_local,
+                    }
+                end
+            end
+            if #rows == 0 then renderer.status("(no usage recorded)"); return end
+            -- R7: 3-level deterministic sort
+            table.sort(rows, function(a, b)
+                if a.cost ~= b.cost     then return a.cost > b.cost end
+                if a.model ~= b.model   then return a.model < b.model end
+                return a.category < b.category
+            end)
+            renderer.status(("session usage detail (total=$%.6f, %d/%d tokens):"):format(
+                total_cost, total_p, total_c))
+            for _, r in ipairs(rows) do
+                io.write(("  %-26s %-18s %3d calls, %6d / %6d tokens, $%.6f%s\n"):format(
+                    r.model, r.category, r.calls, r.prompt, r.completion, r.cost,
+                    r.is_local and "  (local)" or ""))
+            end
+            return
+        end
+        renderer.status("usage: :cost [detail|reset]")
+    end
     -- diff or git failure emits status and skips — never pollutes
     -- context with empty or error noise.
     meta.diff = function(args)