tickmarkr 2.8.6 → 2.8.8

package/dist/run/repair-disposition.d.ts

@@ -4,7 +4,7 @@ import type { GateResult } from "../gates/types.js";
 export type RepairDispositionKind = "fund-repair" | "scope-request" | "authoring" | "none";
 export interface RepairDisposition {
     kind: RepairDispositionKind;
-    /** validated inventory paths only — the sole paths a caller may put in an executable command */
+    /** Queue row 235 (D-2134): validated inventory paths or explicitly declared oracle landings. */
     paths: string[];
     reason: string;
     /** which evidence decided: a gate name, "worker" for a refusal, absent for the ordinary disposition */
@@ -33,7 +33,7 @@ export interface RepairDispositionInput {
     /** task tree plus diff, deleted tracked paths included */
     inventory: Iterable<string>;
 }
-/** Lexing never starts in the middle of a word (same lexer as the daemon's scope-red disposition). */
+/** Queue row 230: tsc's position, numeric or baseline-normalized (#), ends a path; named parentheses stay. */
 export declare const namedPaths: (text: string) => string[];
 /** Binds candidates to the SUPPLIED inventory and nothing else: an exact path, else a unique
  * path-suffix. Anything unbound or bound twice comes back as a non-executable diagnostic. */

package/dist/run/repair-disposition.js

@@ -1,8 +1,8 @@
 import { filesGlob } from "../graph/files-glob.js";
 const failed = (g) => !(g.pass || g.meta?.skipped === true) || g.meta?.infra === true;
 const REFUSAL_RE = /outside|out.of.scope|allowlist|scope expansion|not (?:in|own)|unowned/i;
-/** Lexing never starts in the middle of a word (same lexer as the daemon's scope-red disposition). */
-export const namedPaths = (text) => [...text.matchAll(/(?:^|[\s`'"])((?:[A-Za-z0-9_@.()[\]-]+\/)+[A-Za-z0-9_@.[\]-]+|[A-Za-z0-9_@-]+(?:\.[A-Za-z0-9_-]+)+)(?=$|[\s`'"),:;.!?])/g)]
+/** Queue row 230: tsc's position, numeric or baseline-normalized (#), ends a path; named parentheses stay. */
+export const namedPaths = (text) => [...text.matchAll(/(?:^|[\s`'"])((?:[A-Za-z0-9_@.()[\]-]+\/)+(?:[A-Za-z0-9_@.[\]-]|\([A-Za-z0-9_@.[\]-]*\))+|(?:[A-Za-z0-9_@-]|\([A-Za-z0-9_@.[\]-]*\))+(?:\.(?:[A-Za-z0-9_-]|\([A-Za-z0-9_@.[\]-]*\))+)+)(?=$|[\s`'"),:;.!?]|\((?:\d+|#),\s*(?:\d+|#)\))/g)]
     .map((match) => match[1].replace(/^\.\//, "").replace(/\.$/, ""))
     .filter((path) => !path.split("/").includes(".."));
 /** Binds candidates to the SUPPLIED inventory and nothing else: an exact path, else a unique
@@ -48,12 +48,42 @@ export function classifyRepairDisposition(input) {
         return { paths: r.resolved.filter((path) => !allowed(path)), diagnostics: r.diagnostics };
     };
     const owned = (text) => resolveScopeHints(namedPaths(text), inventory).resolved.some(allowed);
+    // Queue row 235 (D-2134): acceptance prose quotes declarations; only execution and typed evidence
+    // attribute a repair. A misplaced declared landing is authoritative even when it was never written.
+    const acceptanceEvidence = (g) => {
+        if (Array.isArray(g.meta?.oracles)) {
+            const paths = [];
+            const landings = [];
+            for (const oracle of g.meta.oracles) {
+                if (!oracle || typeof oracle !== "object")
+                    continue;
+                if (typeof oracle.output === "string")
+                    paths.push(...namedPaths(oracle.output));
+                if (["absent", "elsewhere", "untitled"].includes(oracle.failure) && typeof oracle.landing === "string" && !allowed(oracle.landing)) {
+                    landings.push(oracle.landing);
+                }
+            }
+            return { paths, landings };
+        }
+        const judgment = g.meta?.judgment;
+        const criteria = Array.isArray(judgment) ? judgment
+            : judgment && typeof judgment === "object" && "criteria" in judgment ? judgment.criteria : undefined;
+        const paths = Array.isArray(criteria) ? criteria.flatMap((row) => row?.met === false && Array.isArray(row.paths) ? row.paths.filter((path) => typeof path === "string") : []) : [];
+        return { paths, landings: [] };
+    };
     const refusal = input.refusalSummary !== undefined && REFUSAL_RE.test(input.refusalSummary)
         ? unowned(input.refusalSummary) : { paths: [], diagnostics: [] };
     // A test gate names where the ASSERTION lives — a detection site, not the code that needs fixing.
     const sites = reds.filter((g) => g.gate === "test").map((g) => unowned(g.details));
     const others = reds.filter((g) => g.gate !== "test");
-    const hints = others.map((g) => unowned(g.details));
+    const evidence = others.map((g) => g.gate === "acceptance" ? acceptanceEvidence(g) : undefined);
+    const hints = others.map((g, i) => {
+        const e = evidence[i];
+        if (!e)
+            return unowned(g.details);
+        const r = resolveScopeHints(e.paths.filter((path) => !allowed(path)), inventory);
+        return { paths: [...e.landings, ...r.resolved.filter((path) => !allowed(path))], diagnostics: r.diagnostics };
+    });
     const diagnostics = [refusal, ...sites, ...hints].flatMap((r) => r.diagnostics)
         .filter((d, i, all) => all.findIndex((o) => o.candidate === d.candidate) === i);
     const uniq = (rs) => [...new Set(rs.flatMap((r) => r.paths))].sort();
@@ -62,7 +92,8 @@ export function classifyRepairDisposition(input) {
             reason: `worker refusal names unowned paths: ${refusal.paths.join(", ")}` };
     }
     const repair = uniq(hints);
-    if (repair.length && !others.some((g) => owned(g.details))) {
+    if (repair.length && !others.some((g, i) => evidence[i]
+        ? resolveScopeHints(evidence[i].paths, inventory).resolved.some(allowed) : owned(g.details))) {
         const review = others.some((g) => g.gate === "review");
         return { kind: review ? "scope-request" : "authoring", paths: repair, source: others[0].gate, diagnostics,
             reason: `files[] repair hint: ${repair.join(", ")}` };

package/dist/run/stall.d.ts

@@ -3,6 +3,7 @@
 export declare function normalizeStallSnapshot(text: string): string;
 export interface StallProgressSample {
     paneText: string;
+    receivedBytes?: number;
     seedSubmitted?: boolean;
     contextTokens?: number;
 }
@@ -102,6 +103,7 @@ export declare const PANE_READ_ROWS = 1000;
  */
 export declare class StallProgressTracker {
     private transcriptRows;
+    private receivedBytes;
     private rawWindowLines;
     private seedSubmitted;
     private contextTokens;

package/dist/run/stall.js

@@ -613,6 +613,7 @@ export const PANE_READ_ROWS = 1000;
  */
 export class StallProgressTracker {
     transcriptRows = 0; // non-empty high-water over the bounded read — the growth signal
+    receivedBytes = 0;
     rawWindowLines = 0; // raw-line high-water — the SATURATION signal (see the getter)
     seedSubmitted = false;
     contextTokens;
@@ -654,6 +655,14 @@ export class StallProgressTracker {
             rowsAdvanced = true;
             this.rowGrowthAt = now; // raw signal — advances even when the report below is suppressed
         }
+        // Queue row 233 (D-2130): print-mode progress is the driver's received-byte count, never a
+        // row high-water over the bounded tail. A driver reporting no count keeps the row measure.
+        const bytes = sample.receivedBytes;
+        let outputAdvanced = rowsAdvanced;
+        if (bytes !== undefined && Number.isFinite(bytes) && bytes >= 0) {
+            outputAdvanced = bytes > this.receivedBytes;
+            this.receivedBytes = Math.max(this.receivedBytes, bytes);
+        }
         let seedAdvanced = false;
         if (sample.seedSubmitted && !this.seedSubmitted) {
             this.seedSubmitted = true;
@@ -674,10 +683,9 @@ export class StallProgressTracker {
         }
         if (tokensAdvanced || seedAdvanced)
             return true;
-        if (rowsAdvanced) {
-            // T1: row growth past the flat-token cap is cosmetic — the paid-work counter has not moved,
-            // so the inactivity window must NOT re-arm on it. A tracker that never sees a token sample
-            // (unmetered adapter) keeps the old row-growth behavior.
+        if (outputAdvanced) {
+            // Queue row 233 (D-2130): both output measures obey the existing flat-token cap;
+            // an unmetered adapter can keep re-arming on output indefinitely, up to its hard deadline.
             if (this.lastTokenGrowthAt !== undefined && now - this.lastTokenGrowthAt >= rowRearmTokenFlatMs)
                 return false;
             return true;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "tickmarkr",
-  "version": "2.8.6",
+  "version": "2.8.8",
   "description": "Spec in, verified work out.",
   "type": "module",
   "license": "MIT",

package/skills/tickmarkr-auto/SKILL.md

@@ -20,16 +20,18 @@ Keep `name: tickmarkr-auto` and the installed links to canonical `skills/` intac
 
 ## Two-tier by default — role check before the loop
 
+`<label>` means the version only in the tickmarkr source repository (`package.json` named `tickmarkr`); elsewhere it means the mission or run, never the installed binary version.
+
 When working in a multi-agent terminal environment (herdr or Orca), decide your role before starting:
 
-- **Orchestrator:** your session was started to execute the mission. Rename your own tab/pane `ORCH · <version>` (short labels: ≤20 chars, `ROLE · token`) and run the loop below.
+- **Orchestrator:** your session was started to execute the mission. Rename your own tab/pane `ORCH · <label>` (short labels: ≤20 chars, `ROLE · token`) and run the loop below.
 - **Supervisor with a live orchestrator:** do not start a second run. Relay the mission to the existing orchestrator with a [verified handoff](#verified-handoffs-agent-to-agent-messaging), then supervise it as OVERSEER.
-- **Primary session without an orchestrator:** rename your own tab `OVERSEER · <version>` and your agent `overseer`, spawn one child orchestration session with your host's launch form, label its tab `ORCH · <version>` and name its agent, give it the mission and these rules verbatim, then supervise it. Do not drive a duplicate single-tier run yourself. Before spawning, confirm any PREVIOUS orchestrator has stood down (monitors stopped, input box empty — dim ghost-text suggestions are UI, not queued input; ANSI-verify before alarming) and close its tab — the journal, records, and ledger hold the story; scrollback is disposable.
-  - **On herdr (`HERDR_ENV=1`)**: **spawning on current herdr is two-step** — the one-shot `agent start --cwd/--tab/--no-focus` form was removed in the herdr CLI redesign and now fails with `unknown option` (OBS-138). First create the pane: `herdr tab create --workspace <ws> --cwd <repo> --label "ORCH · <version>"` (tab create does not steal focus unless `--focus` is passed; parse `result.root_pane.pane_id` from its JSON), then start the agent in it:
+- **Primary session without an orchestrator:** rename your own tab `OVERSEER · <label>` and your agent `overseer`, spawn one child orchestration session with your host's launch form, label its tab `ORCH · <label>` and name its agent, give it the mission and these rules verbatim, then supervise it. Do not drive a duplicate single-tier run yourself. Before spawning, confirm any PREVIOUS orchestrator has stood down (monitors stopped, input box empty — dim ghost-text suggestions are UI, not queued input; ANSI-verify before alarming) and close its tab — the journal, records, and ledger hold the story; scrollback is disposable.
+  - **On herdr (`HERDR_ENV=1`)**: **spawning on current herdr is two-step** — the one-shot `agent start --cwd/--tab/--no-focus` form was removed in the herdr CLI redesign and now fails with `unknown option` (OBS-138). First create the pane: `herdr tab create --workspace <ws> --cwd <repo> --label "ORCH · <label>"` (tab create does not steal focus unless `--focus` is passed; parse `result.root_pane.pane_id` from its JSON), then start the agent in it:
     - **Claude Code:** `herdr agent start orchestrator --kind claude --pane <root-pane-id> -- --permission-mode bypassPermissions`
     - **Codex:** `herdr agent start orchestrator --kind codex --pane <root-pane-id> -- --dangerously-bypass-approvals-and-sandbox` — the unsandboxed flag is REQUIRED, not optional: codex's `workspace-write` sandbox keeps `.git` refs read-only, so a sandboxed orchestrator's `tickmarkr run` dies at integration-branch creation (`git worktree add` cannot lock the ref). Do not downgrade this flag; the herdr pane and repo scope are the containment.
     - **Auxiliary seats run as the CLI's interactive TUI in their visible pane — never headless** (`claude -p` / `codex exec`): headless buffers output until exit so the pane renders idle for the entire run, is blind to SessionStart hook errors (a broken and a fixed hook both return green), and gives a stall watcher no midpoint — silent-time equals lifetime. Headless is for exit-code probes only (a quota check that wants `rc`), never for work anyone must watch.
-  - **On Orca (`TERM_PROGRAM=Orca` and non-empty `ORCA_TERMINAL_HANDLE`)**: spawn it with `orca terminal create --worktree path:<repo> --title "ORCH · <version>" --command "<agent-cmd>" --json` and parse `result.terminal.handle` from the create receipt. **Auxiliary seats run as the CLI's interactive TUI in their visible terminal — never headless**, for the reasons the herdr branch gives.
+  - **On Orca (`TERM_PROGRAM=Orca` and non-empty `ORCA_TERMINAL_HANDLE`)**: spawn it with `orca terminal create --worktree path:<repo> --title "ORCH · <label>" --command "<agent-cmd>" --json` and parse `result.terminal.handle` from the create receipt. **Auxiliary seats run as the CLI's interactive TUI in their visible terminal — never headless**, for the reasons the herdr branch gives.
   - **Auxiliary agents you spawn (consultants, reviewers, scouts) follow the same forms.** Never launch a claude session in plan mode or default permission mode for autonomous work — both stall on per-command approval prompts nobody is watching; claude is always `--permission-mode bypassPermissions` (tickmarkr's own adapter uses exactly this for workers, judges, and consults). A read-only codex consultant may use `--sandbox read-only`; any codex session that must touch git needs the unsandboxed flag above.
 
 With neither herdr nor Orca there is no role to decide and no seat to spawn: follow [On a plain host (neither herdr nor Orca)](#on-a-plain-host-neither-herdr-nor-orca) and run the loop directly, single-tier.
@@ -83,11 +85,11 @@ Run every requested target in order without seeking routine confirmation. Stop o
 
 ## Binary preflight (before compile or run)
 
-Before `tickmarkr compile` or `tickmarkr run`, compare the installed binary against the repository's `package.json` version:
+Before `tickmarkr compile` or `tickmarkr run`:
 
-1. Run `tickmarkr version` (one line, machine-parseable).
-2. Read the `version` field from the repository's `package.json`.
-3. If the binary and repository do not **agree on the entire version** (including the patch; e.g. binary `2.1.0` vs repo `2.1.1`), **stop immediately** and tell the operator to update the global install (`npm i -g tickmarkr@latest`), or to install this repository's build as a REAL COPY — `npm pack`, then `npm i -g ./<tarball>`. Do not compile, plan, or run on hope.
+1. Run `tickmarkr version --check` and **stop immediately on a non-zero exit**.
+2. In a consumer repository, it compares installed agent guidance with this binary's packaged block and skills, never the app's manifest version. Refresh drifted guidance with `tickmarkr init --agent --force`.
+3. Only in the tickmarkr source repository (`package.json` named `tickmarkr`), the check compares its version with the binary. If the binary and repository do not **agree on the entire version** (including the patch; e.g. binary `2.1.0` vs repo `2.1.1`), **stop immediately** and tell the operator to update the global install (`npm i -g tickmarkr@latest`), or to install this repository's build as a REAL COPY — `npm pack`, then `npm i -g ./<tarball>`. Do not compile, plan, or run on hope.
    > ⚠ **Never `npm i -g .` on the repository directory, and never link it.** npm SYMLINKS a directory install, which makes the working tree itself the machine-wide binary: every later build — including a gate's own `npm run build` — silently hot-swaps the CLI for every repository on the machine, with no version change to notice it by. Measured 2026-08-29: a verify build gate rewrote the shared binary while another repository's daemon was mid-run against it, and a positive control that rebuilds at a pre-fix ref would have installed the very defect it was proving fixed, machine-wide (OBS-771). Verify an install by comparing the global and repo **inodes** — they must DIFFER — never by `tickmarkr version`, which cannot go red when nothing is bumped.
 
 A stale binary silently skips daemon gates shipped in newer releases — the v1.38 run exposed this when a global `1.36.0` binary missed the daemon tip-verify gate entirely (OBS-38). Preflight failure is always stop-and-report; never proceed-and-hope.

package/skills/tickmarkr-loop/SKILL.md

@@ -9,16 +9,18 @@ Use this for any spec that `tickmarkr compile` accepts. It is SDD-agnostic: use
 
 ## Two-tier by default — role check before the loop
 
+`<label>` means the version only in the tickmarkr source repository (`package.json` named `tickmarkr`); elsewhere it means the mission or run, never the installed binary version.
+
 When working in a multi-agent terminal environment (herdr or Orca), decide your role before starting:
 
-- **Orchestrator:** your session was started to execute the mission. Rename your own tab/pane `ORCH · <version>` (short labels: ≤20 chars, `ROLE · token`) and run the loop below.
+- **Orchestrator:** your session was started to execute the mission. Rename your own tab/pane `ORCH · <label>` (short labels: ≤20 chars, `ROLE · token`) and run the loop below.
 - **Supervisor with a live orchestrator:** do not start a second run. Relay the mission to the existing orchestrator with a [verified handoff](#verified-handoffs-agent-to-agent-messaging), then supervise it as OVERSEER.
-- **Primary session without an orchestrator:** rename your own tab `OVERSEER · <version>` and your agent `overseer`, spawn one child orchestration session with your host's launch form, label its tab `ORCH · <version>` and name its agent, give it the mission and these rules verbatim, then supervise it. Do not drive a duplicate single-tier run yourself. Before spawning, confirm any PREVIOUS orchestrator has [stood down](#stand-down-mission-end-and-retirement) and close its tab.
-  - **On herdr (`HERDR_ENV=1`)**: **spawning on current herdr is two-step** — the one-shot `agent start --cwd/--tab/--no-focus` form was removed in the herdr CLI redesign and now fails with `unknown option` (OBS-138). First create the pane: `herdr tab create --workspace <ws> --cwd <repo> --label "ORCH · <version>"` (tab create does not steal focus unless `--focus` is passed; parse `result.root_pane.pane_id` from its JSON), then start the agent in it:
+- **Primary session without an orchestrator:** rename your own tab `OVERSEER · <label>` and your agent `overseer`, spawn one child orchestration session with your host's launch form, label its tab `ORCH · <label>` and name its agent, give it the mission and these rules verbatim, then supervise it. Do not drive a duplicate single-tier run yourself. Before spawning, confirm any PREVIOUS orchestrator has [stood down](#stand-down-mission-end-and-retirement) and close its tab.
+  - **On herdr (`HERDR_ENV=1`)**: **spawning on current herdr is two-step** — the one-shot `agent start --cwd/--tab/--no-focus` form was removed in the herdr CLI redesign and now fails with `unknown option` (OBS-138). First create the pane: `herdr tab create --workspace <ws> --cwd <repo> --label "ORCH · <label>"` (tab create does not steal focus unless `--focus` is passed; parse `result.root_pane.pane_id` from its JSON), then start the agent in it:
     - **Claude Code:** `herdr agent start orchestrator --kind claude --pane <root-pane-id> -- --permission-mode bypassPermissions`
     - **Codex:** `herdr agent start orchestrator --kind codex --pane <root-pane-id> -- --dangerously-bypass-approvals-and-sandbox` — the unsandboxed flag is REQUIRED, not optional: codex's `workspace-write` sandbox keeps `.git` refs read-only, so a sandboxed orchestrator's `tickmarkr run` dies at integration-branch creation (`git worktree add` cannot lock the ref). Do not downgrade this flag; the herdr pane and repo scope are the containment.
     - **Auxiliary seats run as the CLI's interactive TUI in their visible pane — never headless** (`claude -p` / `codex exec`): headless buffers output until exit so the pane renders idle for the entire run, is blind to SessionStart hook errors (a broken and a fixed hook both return green), and gives a stall watcher no midpoint — silent-time equals lifetime. Headless is for exit-code probes only (a quota check that wants `rc`), never for work anyone must watch.
-  - **On Orca (`TERM_PROGRAM=Orca` and non-empty `ORCA_TERMINAL_HANDLE`)**: spawn it with `orca terminal create --worktree path:<repo> --title "ORCH · <version>" --command "<agent-cmd>" --json` and parse `result.terminal.handle` from the create receipt. **Auxiliary seats run as the CLI's interactive TUI in their visible terminal — never headless**, for the reasons the herdr branch gives.
+  - **On Orca (`TERM_PROGRAM=Orca` and non-empty `ORCA_TERMINAL_HANDLE`)**: spawn it with `orca terminal create --worktree path:<repo> --title "ORCH · <label>" --command "<agent-cmd>" --json` and parse `result.terminal.handle` from the create receipt. **Auxiliary seats run as the CLI's interactive TUI in their visible terminal — never headless**, for the reasons the herdr branch gives.
   - **Auxiliary agents you spawn (consultants, reviewers, scouts) follow the same forms.** Never launch a claude session in plan mode or default permission mode for autonomous work — both stall on per-command approval prompts nobody is watching; claude is always `--permission-mode bypassPermissions` (tickmarkr's own adapter uses exactly this for workers, judges, and consults). A read-only codex consultant may use `--sandbox read-only`; any codex session that must touch git needs the unsandboxed flag above.
 
 With neither herdr nor Orca there is no role to decide and no seat to spawn: follow [On a plain host (neither herdr nor Orca)](#on-a-plain-host-neither-herdr-nor-orca) and run the loop directly, single-tier.
@@ -69,11 +71,11 @@ Proceed through the loop without seeking routine confirmation. Stop only for a b
 
 ## Binary preflight (before compile or run)
 
-Before `tickmarkr compile` or `tickmarkr run`, compare the installed binary against the repository's `package.json` version:
+Before `tickmarkr compile` or `tickmarkr run`:
 
-1. Run `tickmarkr version` (one line, machine-parseable).
-2. Read the `version` field from the repository's `package.json`.
-3. If the binary and repository do not **agree on the entire version** (including the patch; e.g. binary `2.1.0` vs repo `2.1.1`), **stop immediately** and tell the operator to update the global install (`npm i -g tickmarkr@latest`), or to install this repository's build as a REAL COPY — `npm pack`, then `npm i -g ./<tarball>`. Do not compile, plan, or run on hope.
+1. Run `tickmarkr version --check` and **stop immediately on a non-zero exit**.
+2. In a consumer repository, it compares installed agent guidance with this binary's packaged block and skills, never the app's manifest version. Refresh drifted guidance with `tickmarkr init --agent --force`.
+3. Only in the tickmarkr source repository (`package.json` named `tickmarkr`), the check compares its version with the binary. If the binary and repository do not **agree on the entire version** (including the patch; e.g. binary `2.1.0` vs repo `2.1.1`), **stop immediately** and tell the operator to update the global install (`npm i -g tickmarkr@latest`), or to install this repository's build as a REAL COPY — `npm pack`, then `npm i -g ./<tarball>`. Do not compile, plan, or run on hope.
    > ⚠ **Never `npm i -g .` on the repository directory, and never link it.** npm SYMLINKS a directory install, which makes the working tree itself the machine-wide binary: every later build — including a gate's own `npm run build` — silently hot-swaps the CLI for every repository on the machine, with no version change to notice it by. Measured 2026-08-29: a verify build gate rewrote the shared binary while another repository's daemon was mid-run against it, and a positive control that rebuilds at a pre-fix ref would have installed the very defect it was proving fixed, machine-wide (OBS-771). Verify an install by comparing the global and repo **inodes** — they must DIFFER — never by `tickmarkr version`, which cannot go red when nothing is bumped.
 
 A stale binary silently skips daemon gates shipped in newer releases — the v1.38 run exposed this when a global `1.36.0` binary missed the daemon tip-verify gate entirely (OBS-38). Preflight failure is always stop-and-report; never proceed-and-hope.

package/skills/tickmarkr-overseer/SKILL.md

@@ -5,6 +5,8 @@ description: "Use when the user asks to oversee/supervise/babysit an autonomous
 
 # Overseer (tickmarkr)
 
+`<label>` means the version only in the tickmarkr source repository (`package.json` named `tickmarkr`); elsewhere it means the mission or run, never the installed binary version.
+
 Become the OVERSEER for this workspace or session. Do no heavy work directly — build and supervise a two-tier
 hierarchy of VISIBLE agents (you → orchestrator → tickmarkr's own worker fleet), and route human decisions
 to the user with evidence.
@@ -203,15 +205,15 @@ replaces the two-tier setup.
      **Live tab labels (standing operator rule, 2026-07-12):** on every decision or state change (role
      handoff, task done/merged, run end) rename the affected tabs — and keep labels SHORT: the role as the
      main name plus at most ONE hot-state token. Vocabulary: ORCH carries the milestone and progress
-     fraction (`ORCH · v1.19 4/5`, updated on every task-done); tickmarkr opens ONE TAB PER TASK, labelled
+     fraction (`ORCH · <label> 4/5`, updated on every task-done); tickmarkr opens ONE TAB PER TASK, labelled
      with the task id and holding that task's worker plus its judge/review/consult panes (tickmarkr
      updates it). Never long context strings or ✓-chains.
-   - **On Orca (`TERM_PROGRAM=Orca` and non-empty `ORCA_TERMINAL_HANDLE`)**: Do not load the `herdr` skill and do not map a Herdr workspace. Work from the current Orca terminal context. Keep the overseer in the launching terminal, name it in the same act — `orca terminal rename --terminal "$ORCA_TERMINAL_HANDLE" --title "OVERSEER · <version>"` (tab title; see the seat-name law under Seat-spawn recipes) — and inspect terminals with `orca terminal list --json`. Launch the daemon detached at the orchestrator terminal's recorded handle using the detached launch below — no pane of its own; its self-placed watch board stays in the ORCH tab.
+   - **On Orca (`TERM_PROGRAM=Orca` and non-empty `ORCA_TERMINAL_HANDLE`)**: Do not load the `herdr` skill and do not map a Herdr workspace. Work from the current Orca terminal context. Keep the overseer in the launching terminal, name it in the same act — `orca terminal rename --terminal "$ORCA_TERMINAL_HANDLE" --title "OVERSEER · <label>"` (tab title; see the seat-name law under Seat-spawn recipes) — and inspect terminals with `orca terminal list --json`. Launch the daemon detached at the orchestrator terminal's recorded handle using the detached launch below — no pane of its own; its self-placed watch board stays in the ORCH tab.
 2. **Orchestrator**: First consume a successful `tickmarkr fleet --pick consult` in the seat repository, per the Fleet selection contract below; then launch the returned adapter/model with your agent host.
-   - **On herdr (`HERDR_ENV=1`)**: Spawning on current herdr is two-step — the one-shot `agent start --cwd` form was removed in the herdr CLI redesign and now fails with `unknown option` (OBS-138): first create the pane with `herdr tab create --workspace <ws> --cwd <repo> --label "ORCH · <version>"` and parse `result.root_pane.pane_id` from its JSON, then start the agent in it. For Claude Code, use `herdr agent start orchestrator --kind claude --pane <root-pane-id> -- --permission-mode bypassPermissions` (append `--model <picked-model>` after the `--` from the Fleet receipt). For Codex, use `herdr agent start orchestrator --kind codex --pane <root-pane-id> -- --dangerously-bypass-approvals-and-sandbox` (add `--model <picked-model>` from the Fleet receipt). The unsandboxed flag is REQUIRED: codex's `workspace-write` sandbox keeps `.git` refs read-only, so a sandboxed orchestrator's `tickmarkr run` dies at integration-branch creation — do not downgrade it. Workers you never spawn — tickmarkr spawns its own visible worker panes. Auxiliary agents you do spawn (consultants, reviewers, scouts) follow the same forms: never launch a claude session in plan mode or default permission mode for autonomous work — both stall on per-command approval prompts nobody is watching; claude is always `--permission-mode bypassPermissions --settings '{"promptSuggestionEnabled":false}'`. **For a codex consultant, use `-a never --sandbox workspace-write` — NOT `--sandbox read-only`.** ⚠ **`--sandbox read-only` CONTRADICTS this skill's own completion protocol and will hang the seat.** Every seat you spawn is told to deliver an ARTIFACT ending in a terminal MARKER, because that is the only completion signal the artifact watcher can key on (`done` is turn end). A read-only sandbox cannot write that artifact, so codex blocks on `Would you like to make the following edits?` for its OWN report — and the report exists ONLY in the pending edit, so abandoning the prompt destroys the work rather than merely delaying it. Measured 2026-08-28: a consultant spawned `--sandbox read-only` finished a 14,604-byte verdict, sat blocked on the write, and the operator saw the prompt before the supervising tier did. `read-only` is correct ONLY for a seat that writes nothing at all — which, under the artifact+marker rule, is no seat this skill tells you to spawn. When the prompt does appear, answer **"Yes, and don't ask again for these files"** rather than plain yes: plain yes re-blocks on the next write of the same file. **That `--settings` pair is not cosmetic and it is not optional:** claude-code's AUTOSUGGEST renders context-plausible ghost text into an idle seat's prompt line that is BYTE-IDENTICAL to a typed draft in text-format reads (OBS-482), so a supervising tier cannot tell a seat's own unsent work from a rendering artifact without `agent read --format ansi`. Turning the suggester off at spawn removes the ambiguity at its source instead of paying for the discrimination at every read. Verified against the shipped binary: `claude --settings '{"promptSuggestionEnabled":false}' -p …` exits 0 with a real response, and the key appears in the binary's own settings schema. **For kimi, pass `-y`** (`herdr agent start <name> --kind kimi --pane <id> -- -y`) — the adapter already launches its own workers that way (`src/adapters/kimi.ts:204`), and a kimi seat spawned without it sits on an approval prompt having done nothing. **Herdr cannot see that state**: it reports a kimi pane as `agent_status: working` with `screen_detection_skipped: true` while the prompt is up, so the BLOCKED-STATE watcher below is blind on this vendor and the spawn flag is the ONLY control. Every vendor you spawn needs its auto-approve form named here; a vendor absent from this list is a seat that will hang.
+   - **On herdr (`HERDR_ENV=1`)**: Spawning on current herdr is two-step — the one-shot `agent start --cwd` form was removed in the herdr CLI redesign and now fails with `unknown option` (OBS-138): first create the pane with `herdr tab create --workspace <ws> --cwd <repo> --label "ORCH · <label>"` and parse `result.root_pane.pane_id` from its JSON, then start the agent in it. For Claude Code, use `herdr agent start orchestrator --kind claude --pane <root-pane-id> -- --permission-mode bypassPermissions` (append `--model <picked-model>` after the `--` from the Fleet receipt). For Codex, use `herdr agent start orchestrator --kind codex --pane <root-pane-id> -- --dangerously-bypass-approvals-and-sandbox` (add `--model <picked-model>` from the Fleet receipt). The unsandboxed flag is REQUIRED: codex's `workspace-write` sandbox keeps `.git` refs read-only, so a sandboxed orchestrator's `tickmarkr run` dies at integration-branch creation — do not downgrade it. Workers you never spawn — tickmarkr spawns its own visible worker panes. Auxiliary agents you do spawn (consultants, reviewers, scouts) follow the same forms: never launch a claude session in plan mode or default permission mode for autonomous work — both stall on per-command approval prompts nobody is watching; claude is always `--permission-mode bypassPermissions --settings '{"promptSuggestionEnabled":false}'`. **For a codex consultant, use `-a never --sandbox workspace-write` — NOT `--sandbox read-only`.** ⚠ **`--sandbox read-only` CONTRADICTS this skill's own completion protocol and will hang the seat.** Every seat you spawn is told to deliver an ARTIFACT ending in a terminal MARKER, because that is the only completion signal the artifact watcher can key on (`done` is turn end). A read-only sandbox cannot write that artifact, so codex blocks on `Would you like to make the following edits?` for its OWN report — and the report exists ONLY in the pending edit, so abandoning the prompt destroys the work rather than merely delaying it. Measured 2026-08-28: a consultant spawned `--sandbox read-only` finished a 14,604-byte verdict, sat blocked on the write, and the operator saw the prompt before the supervising tier did. `read-only` is correct ONLY for a seat that writes nothing at all — which, under the artifact+marker rule, is no seat this skill tells you to spawn. When the prompt does appear, answer **"Yes, and don't ask again for these files"** rather than plain yes: plain yes re-blocks on the next write of the same file. **That `--settings` pair is not cosmetic and it is not optional:** claude-code's AUTOSUGGEST renders context-plausible ghost text into an idle seat's prompt line that is BYTE-IDENTICAL to a typed draft in text-format reads (OBS-482), so a supervising tier cannot tell a seat's own unsent work from a rendering artifact without `agent read --format ansi`. Turning the suggester off at spawn removes the ambiguity at its source instead of paying for the discrimination at every read. Verified against the shipped binary: `claude --settings '{"promptSuggestionEnabled":false}' -p …` exits 0 with a real response, and the key appears in the binary's own settings schema. **For kimi, pass `-y`** (`herdr agent start <name> --kind kimi --pane <id> -- -y`) — the adapter already launches its own workers that way (`src/adapters/kimi.ts:204`), and a kimi seat spawned without it sits on an approval prompt having done nothing. **Herdr cannot see that state**: it reports a kimi pane as `agent_status: working` with `screen_detection_skipped: true` while the prompt is up, so the BLOCKED-STATE watcher below is blind on this vendor and the spawn flag is the ONLY control. Every vendor you spawn needs its auto-approve form named here; a vendor absent from this list is a seat that will hang.
    - **On Orca (`TERM_PROGRAM=Orca` and non-empty `ORCA_TERMINAL_HANDLE`)**: Seats spawn with `orca terminal create` on a path worktree selector with a command:
-     `orca terminal create --worktree path:<repo> --title "ORCH · <version>" --command "<agent-cmd>" --json`
-     For Claude Code: `orca terminal create --worktree path:<repo> --title "ORCH · <version>" --command "claude --model <picked-model> --permission-mode bypassPermissions --settings '{\"promptSuggestionEnabled\":false}'" --json`. For Codex: `orca terminal create --worktree path:<repo> --title "ORCH · <version>" --command "codex --model <picked-model> --dangerously-bypass-approvals-and-sandbox" --json`. Parse `result.terminal.handle` from the create receipt.
+     `orca terminal create --worktree path:<repo> --title "ORCH · <label>" --command "<agent-cmd>" --json`
+     For Claude Code: `orca terminal create --worktree path:<repo> --title "ORCH · <label>" --command "claude --model <picked-model> --permission-mode bypassPermissions --settings '{\"promptSuggestionEnabled\":false}'" --json`. For Codex: `orca terminal create --worktree path:<repo> --title "ORCH · <label>" --command "codex --model <picked-model> --dangerously-bypass-approvals-and-sandbox" --json`. Parse `result.terminal.handle` from the create receipt.
    Require the orchestrator to use the host-owned daemon and detached-beat lifecycle below for both run and resume, and to report the lock pid ancestry check.
 3. **Standing instructions travel as a brief FILE, never as pane text** — PTY input truncates at ~1024B and a
    truncated brief silently drops policy. Write the full brief to `<repo>/.tickmarkr/overseer/ORCH-BRIEF.md`
@@ -221,7 +223,7 @@ replaces the two-tier setup.
    - **On Orca (`TERM_PROGRAM=Orca` and non-empty `ORCA_TERMINAL_HANDLE`)**: Briefs travel as a file announced by `terminal send` with enter and a submit wait whose receipt is read for observed submission rather than input acceptance:
      `orca terminal send --terminal <handle> --text "Read .tickmarkr/overseer/ORCH-BRIEF.md and follow it exactly." --enter --wait-submit 15 --json`
      Read `result.send.prompt.stages` and treat the brief as delivered only when a `turn_started` stage is present; `accepted: true` proves input acceptance, not a started turn; never resend on `accepted` alone.
-   The brief MUST contain: the mission, the five-tab canon (on herdr) or seat layout (on Orca), the pane mechanics below, rules 1–2, the GSD-leg rules
+   The brief MUST contain: **Seat hygiene — in a multi-tier session the tiers clear each other at 50% context or more after writing a handoff; a new spec or run is a mission boundary, so hand off and clear before the next mission; copy this duty into every brief and handoff; on a single-tier plain host write your handoff and ask the operator to clear the session**, then the mission, the five-tab canon (on herdr) or seat layout (on Orca), the pane mechanics below, rules 1–2, the GSD-leg rules
    (below) whenever the mission dispatches `/gsd:*` legs, and require a verbatim one-sentence
    acknowledgment of the human-checkpoint rule before anything is dispatched.
    *method guards* in that brief (how to know a thing, not what is true of this spec), and deleting them
@@ -329,7 +331,7 @@ Kimi `-y`, artifact plus terminal marker, and verified file-brief delivery. Flee
   a `git worktree add` the overseer made for a spec branch, a throwaway clone — `terminal create` hangs and
   returns "Timed out waiting for terminal handle after creation". The lawful form for such a checkout is a
   terminal in a managed worktree with a cwd change in the command:
-  `orca terminal create --worktree path:<managed-repo> --title "ORCH · <version>" --command "cd <checkout> && <agent-cmd>" --json`.
+  `orca terminal create --worktree path:<managed-repo> --title "ORCH · <label>" --command "cd <checkout> && <agent-cmd>" --json`.
   **An Orca-hosted run's checkout is created by Orca** (`orca worktree create --repo id:<repo> --base-branch <spec-branch>`),
   never by `git worktree add`: the orca driver asks `orca worktree current` from inside every task worktree and refuses
   (`selector_not_found`) on an unmanaged checkout — every task fails pre-launch and the board cannot place (OBS-1061,
@@ -353,7 +355,7 @@ Kimi `-y`, artifact plus terminal marker, and verified file-brief delivery. Flee
   Code's turn summary, Codex's last prompt) and is NOT the seat's name. This is vendor-neutral by
   construction — no per-vendor env var, no agent command, no launch-form ritual — which is what makes
   it the same law as Herdr's tab label. So: **the overseer names its OWN seat at Setup 1** with
-  `orca terminal rename --terminal "$ORCA_TERMINAL_HANDLE" --title "OVERSEER · <version>"` (the seat
+  `orca terminal rename --terminal "$ORCA_TERMINAL_HANDLE" --title "OVERSEER · <label>"` (the seat
   it was invoked in is the one seat nobody else launches), every seat it spawns is named by
   `create --title`, and a live-label change (progress fraction, hot-state token) is `rename` again.
   **Prove a name by the field that holds:** `tabs[].title` in