tickmarkr 1.89.0 → 1.90.0

package/dist/run/merge.d.ts

@@ -1,4 +1,5 @@
 import type { TickmarkrConfig } from "../config/config.js";
+import { type Baseline } from "../gates/baseline.js";
 export interface TipVerifyResult {
     gate: string;
     cmd: string;
@@ -7,6 +8,8 @@ export interface TipVerifyResult {
     fingerprints: string[];
     details: string;
     artifact?: string;
+    /** Q121s: nonzero exit whose failures are ALL baseline-recorded — forgiven exactly as the battery forgives. */
+    forgiven?: boolean;
 }
 export declare function integrationBranch(cfg: TickmarkrConfig, runId: string): string;
 export declare function ensureIntegration(repo: string, branch: string, baseRef: string): Promise<string>;
@@ -19,4 +22,4 @@ export declare function mergeTask(intWt: string, taskBranch: string, message: st
         branchTip: string;
     };
 }>;
-export declare function verifyIntegrationTip(intWt: string, commands: Record<string, string>, runDir: string): Promise<TipVerifyResult[]>;
+export declare function verifyIntegrationTip(intWt: string, commands: Record<string, string>, runDir: string, baseline?: Baseline): Promise<TipVerifyResult[]>;

package/dist/run/merge.js

@@ -1,7 +1,7 @@
 import { existsSync, writeFileSync } from "node:fs";
 import { join } from "node:path";
 import { shq } from "../adapters/types.js";
-import { fingerprint } from "../gates/baseline.js";
+import { classifyFailureOutput, fingerprint, freshFailures } from "../gates/baseline.js";
 import { tickmarkrDir } from "../graph/graph.js";
 import { gitHead, linkNodeModules, resolveIntegrationBranch, sh, shGit, shGitOk, WORKTREES_DIR } from "./git.js";
 export function integrationBranch(cfg, runId) {
@@ -44,22 +44,41 @@ export async function mergeTask(intWt, taskBranch, message, gatedCommit) {
     await shGit("git merge --abort", intWt);
     return { ok: false, conflict };
 }
-// OBS-34: strict exit-code verify on the integration tip — no baseline forgiveness.
-export async function verifyIntegrationTip(intWt, commands, runDir) {
+// OBS-34 ruled strict exit-code verify; Q121s (TRIAL T-OBS-1) narrows it: a red whose failure
+// fingerprints are ALL recorded in the run's baseline is forgiven with the battery's own math
+// (freshFailures) — on any repo whose main carries a pre-existing red, strict verify made a green
+// terminus unreachable by construction and misattributed the red to the last merged task.
+// Fail-closed edges kept: no baseline → strict; baseline green for that gate → strict; any fresh
+// fingerprint, or output with no recognizable failure shape, → failed.
+export async function verifyIntegrationTip(intWt, commands, runDir, baseline) {
     const results = [];
     for (const [gate, cmd] of Object.entries(commands)) {
         const r = await sh(cmd, intWt);
         const raw = r.stdout + "\n" + r.stderr;
-        const artifact = r.code !== 0 ? join(runDir, `tip-verify-${gate}.log`) : undefined;
+        const stripped = raw.split(intWt).join("");
+        const entry = baseline?.commands[gate];
+        const { failing, unreadable } = freshFailures(entry, stripped);
+        // `?? 1` is the battery's own default (baseline.ts compareToBaseline): an exitCode-less legacy
+        // entry reads as red-at-baseline there, so it must read the same here or old baselines silently
+        // lose forgiveness. Battery parity on the infra rule too (T9): infrastructure-only output means
+        // the runner never completed a suite — nothing was verified, so nothing is forgivable, however
+        // familiar its fingerprints. Stricter-than-battery edge kept: unreadable output never forgives.
+        const forgiven = r.code !== 0 && entry !== undefined && (entry.exitCode ?? 1) !== 0
+            && failing.length === 0 && !unreadable && classifyFailureOutput(stripped) !== "infra";
+        const pass = r.code === 0 || forgiven;
+        const artifact = pass ? undefined : join(runDir, `tip-verify-${gate}.log`);
         if (artifact)
             writeFileSync(artifact, raw);
         results.push({
             gate,
             cmd,
-            pass: r.code === 0,
+            pass,
             exitCode: r.code,
-            fingerprints: r.code !== 0 ? fingerprint(raw.split(intWt).join("")) : [],
-            details: r.code === 0 ? "exit 0" : `exit ${r.code}`,
+            fingerprints: r.code !== 0 ? fingerprint(stripped) : [],
+            details: r.code === 0 ? "exit 0"
+                : forgiven ? `exit ${r.code} but only baseline-recorded failures (forgiven vs baseline)`
+                    : `exit ${r.code}`,
+            ...(forgiven ? { forgiven: true } : {}),
             ...(artifact ? { artifact } : {}),
         });
     }

package/fixtures/authoring-lints/01-awk-range-self-pass.spec.md

@@ -0,0 +1,12 @@
+<!-- tickmarkr:spec -->
+<!-- provenance: v1.89 Finding 0 awk-range self-pass certified only the task heading -->
+
+## T1: Clean leading task
+- goal: Keep the first extracted task deliberately clean
+- acceptance:
+  - judge: the first task remains ordinary canary data
+
+## T2: Finding after the range start
+- goal: Consume T28 classification without declaring the dependency
+- acceptance:
+  - judge: classification controls the conclusion

package/fixtures/authoring-lints/02-judge-text-key-miss.spec.md

@@ -0,0 +1,7 @@
+<!-- tickmarkr:spec -->
+<!-- provenance: v1.89 judge-text extractor read judge instead of the compiled text key -->
+
+## T1: Typed judge finding
+- goal: Put the finding only in a typed judge item
+- acceptance:
+  - judge: cite the audit before accepting the task

package/fixtures/authoring-lints/03-c1-t41-rendered-observable.spec.md

@@ -0,0 +1,8 @@
+<!-- tickmarkr:spec -->
+<!-- provenance: v1.89 T41 amendment 10422761 repaired the stale 52/64 asserting-test catch-22 -->
+
+## T41: Render the normalized numerator
+- goal: Change the stale demo gate numerator to its normalized value
+- files: src/tui/cockpit/derive.ts, tests/compile/authoring-lints.test.ts
+- acceptance:
+  - judge: the tracked demo renders `52/64` after skipped outcomes leave the pass numerator

package/fixtures/authoring-lints/04-c1-t24-named-file.spec.md

@@ -0,0 +1,8 @@
+<!-- tickmarkr:spec -->
+<!-- provenance: v1.89 T24 amendment 46e744bf repaired the vacuous-grace asserting-test catch-22 -->
+
+## T24: Replace the vacuous grace oracle
+- goal: Replace the helper-only grace proof with the production daemon oracle
+- files: src/run/daemon.ts, tests/compile/authoring-lints.test.ts
+- acceptance:
+  - judge: `tests/run/worktree-evidence.test.ts` carries the production runDaemon differential

package/fixtures/authoring-lints/05-c2-t24-t28-dep-inversion.spec.md

@@ -0,0 +1,7 @@
+<!-- tickmarkr:spec -->
+<!-- provenance: v1.89 T24-T28 inversion required T24 to build its undeclared successor machinery -->
+
+## T24: Inverted classification consumer
+- goal: Use T28 flat CPU classification before that task can run
+- acceptance:
+  - judge: the classification permits the liveness conclusion

package/fixtures/authoring-lints/06-c2-denumbered-coupling.spec.md

@@ -0,0 +1,7 @@
+<!-- tickmarkr:spec -->
+<!-- provenance: v1.89 T26 de-numbered T28 coupling hid the same dependency inversion -->
+
+## T26: Coupling without a task token
+- goal: Conclude only when whose independent flat CPU classification permits the result
+- acceptance:
+  - judge: the later classification changes the conclusion

package/fixtures/authoring-lints/07-c3a-t41-line-count-proxy.spec.md

@@ -0,0 +1,7 @@
+<!-- tickmarkr:spec -->
+<!-- provenance: v1.89 T41 physical-line proxy pressured derive wiring into code golf -->
+
+## T41: Bound structure by a proxy
+- goal: Wire the normalized outcome through the projection
+- acceptance:
+  - judge: the physical line count is not greater than the base count

package/fixtures/authoring-lints/08-c3b-t41-governance-referent.spec.md

@@ -0,0 +1,7 @@
+<!-- tickmarkr:spec -->
+<!-- provenance: v1.89 T41 the-audit referent left asserting-test ownership outside the spec -->
+
+## T41: Delegate scope to governance prose
+- goal: Own the projection surface
+- acceptance:
+  - judge: include every asserting test exposed by the audit

package/fixtures/authoring-lints/09-c4-universals-without-pointer.spec.md

@@ -0,0 +1,7 @@
+<!-- tickmarkr:spec -->
+<!-- provenance: v1.89 T26-T28 open universals omitted the closed enumeration pointer -->
+
+## T28: Open prohibition surface
+- goal: Keep liveness evidence fail-open
+- acceptance:
+  - judge: no pane may supply a liveness conclusion, and no capture can become authored evidence

package/fixtures/authoring-lints/10-c5-t34-conjunct-flood.spec.md

@@ -0,0 +1,7 @@
+<!-- tickmarkr:spec -->
+<!-- provenance: v1.89 T34 criterion 3 flooded one title with independently failing conjuncts -->
+
+## T34: Many behaviors under one title
+- goal: Make criterion failure attributable
+- acceptance:
+  - judge: the driver records contact and preserves provenance while worktree evidence stays authoritative, then failed delivery remains distinct plus journal output stays stable with the same task status

package/fixtures/authoring-lints/11-c6-t34-q3-q9-q20-bundle.spec.md

@@ -0,0 +1,7 @@
+<!-- tickmarkr:spec -->
+<!-- provenance: v1.89 T34 bundled Q3-Q9-Q20 and reran all three when its weakest concern failed -->
+
+## T34: Bundle three queue concerns
+- goal: Subsume Q3, Q9 and Q20 in one implementation task
+- acceptance:
+  - judge: the combined implementation satisfies its bundled concerns

package/fixtures/authoring-lints/12-c7-t24-prose-seam.spec.md

@@ -0,0 +1,8 @@
+<!-- tickmarkr:spec -->
+<!-- provenance: v1.89 T24 production-driver prose seam named no existing exported boundary -->
+
+## T24: Govern the daemon through prose
+- goal: No pane text may supply liveness; enforce the rule through the production driver seam
+- files: src/run/daemon.ts
+- acceptance:
+  - judge: the daemon conclusion follows only independent evidence

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "tickmarkr",
-  "version": "1.89.0",
+  "version": "1.90.0",
   "description": "Spec in, verified work out.",
   "type": "module",
   "license": "MIT",

package/skills/tickmarkr-overseer/SKILL.md

@@ -228,6 +228,11 @@ number — an unmeasured budget is not a small budget.
   sent` reports a delivery that never happened. Never chain `wait && act` or trust a send's exit status —
   confirm the pane exists and read it back. An orchestrator's pane can vanish mid-mission without any event
   reaching you; the first symptom is a dispatch into nothing.
+- **Renaming a live agent kills every watcher keyed on the old NAME.** herdr resolves names live,
+  so after `herdr agent rename` the old name stops existing and any `agent wait <old-name>` or
+  name-keyed poll script exits with `agent_not_running` — an exit shaped exactly like a real wake.
+  Re-arm name-keyed watchers in the same act as the rename; file-keyed artifact watchers are
+  unaffected (one more reason to prefer them).
 - Stale typed input is unclearable via CLI — supersede it:
   `pane run "<-- disregard everything before this arrow (stale draft). ACTUAL: <message>"`.
 

package/skills/tickmarkr-overseer/scripts/watch-contamination.sh

@@ -29,7 +29,10 @@ STATE="$(dirname "$J")/.contamination.seen"
 INFRA_RE='vitest-worker|Timeout calling|JS heap out of memory|ENOMEM|EAGAIN|spawn ENOENT|Killed|SIGKILL'
 
 load1() { uptime | sed 's/.*load averages*: *//' | awk '{print $1}' | tr -d ','; }
-vitest_n() { pgrep -f vitest 2>/dev/null | wc -l | tr -d ' '; }
+# Count real vitest runners only. `pgrep -f vitest` over-counts by an order: it matches the WORD
+# in any argv — worker prompts, VITEST_MAX_FORKS in shell strings (measured 7 vs 2 real, 2026-08-10).
+# [v] keeps the pattern from matching its own grep. Forked workers retitle to "node (vitest N)" (comm).
+vitest_n() { ps -axo comm,command 2>/dev/null | grep -Ec 'node_modules/(\.bin/)?[v]itest|[v]itest/dist/|\([v]itest [0-9]+\)'; }
 
 # Seed the seen-set so we wake on what happens NEXT, not on history already ruled on.
 if [ -f "$J" ]; then

package/skills/tickmarkr-overseer/scripts/watch-pending-input.sh

@@ -29,6 +29,16 @@ status_of() {
 # The rendered prompt line is the condition itself. `agent_status` is the harness's OPINION about the
 # seat and fails in both directions (a wedged worker has reported `idle`, a working one `done`), so the
 # text in the box is the stronger signal — it is what will not run.
+# GHOST DISCRIMINATOR (OBS-482 addendum 34): claude-code AUTOSUGGEST renders its suggestion on the
+# prompt line wrapped in SGR dim — sampled live 2026-08-11: `❯ \e[0m\e[2mcheck T76 progress\e[0m`.
+# Typed text renders default. A dim-wrapped pending line is a GHOST (rendering, not state): skip it —
+# no wake, no streak. Negative scope: sampled on one theme/version; a theme rendering typed input dim
+# would blind this check TOWARD silence, so a wake is still authoritative but its absence is not proof.
+is_ghost() {
+  herdr agent read "$TARGET" --source visible --lines 14 --format ansi 2>/dev/null \
+    | grep -F -- "❯" | grep -F -- "$1" | head -1 | grep -q "$(printf '\033')\[2m"
+}
+
 pending_text() {
   herdr agent read "$TARGET" --source visible --lines 14 2>/dev/null \
     | sed -n 's/^[[:space:]]*❯[[:space:]]*//p' | head -1 \
@@ -126,6 +136,7 @@ while [ "$elapsed" -lt "$CAP" ]; do
   case "$S" in
     idle|done)
       T=$(pending_text)
+      if [ -n "$T" ] && is_ghost "$T"; then T=""; fi
       if [ -n "$T" ]; then
         streak=$((streak + 1))
         if [ "$streak" -ge "$CONFIRM" ]; then
@@ -159,6 +170,13 @@ while [ "$elapsed" -lt "$CAP" ]; do
           fi
           echo "PENDING_INPUT $TARGET status=$S sustained=$((streak * POLL))s"
           echo "  unsubmitted: $T"
+          # GHOST WARNING (OBS-482 addendum 33): claude-code AUTOSUGGEST renders context-plausible
+          # ghost text on this same prompt line, and in text-format reads it is BYTE-IDENTICAL to a
+          # typed draft. Ghost text renders dim/grey; typed text renders default. Emit the styled
+          # bytes so the reader can discriminate before treating this as a draft.
+          echo "  styling evidence (dim/grey SGR around the text = autosuggest ghost, NOT a draft):"
+          herdr agent read "$TARGET" --source visible --lines 14 --format ansi 2>/dev/null \
+            | grep -F -- "$T" | head -2 | cat -v | sed 's/^/    /'
           echo "  the seat is idle and holding live work in its prompt — supersede it, do not re-send:"
           echo "  herdr agent prompt $TARGET \" <-- disregard everything before this arrow (stale draft). ACTUAL: …\""
           # Name the ACTUAL reason we fell through. Saying "budget exhausted" when the cause was the