worktrust 0.8.8 → 0.8.9

package/README.md

@@ -173,6 +173,7 @@ nothing it does not:
 | `unconfirmed` | checks the stretch ran whose outcome no reader could see (Codex's code mode prints no exit code; a background terminal), per kind: counted, never as a pass (0.8.5) |
 | `autonomy`, `authorship` | the permission mode the AI app ran in, on one scale (ask, edits, plan, auto, full), how often it changed and the turns in plan mode; of the stretch's commits, how many name an AI as co-author (the trailer is read here, its names never leave) (0.8.7) |
 | `framing`, `quality`, `risk`, `tool_mix`, `reads`, `context_files` | how the work was framed (your turns before the agent's first action, a plan first or not), checked (a check passing after the last change, the failures before a check passed, the change looked at before delivery; security scans count as checks), risked (destructive commands proposed, refused, run), which kinds of tools were used, files read again unchanged, and which of the AI app's own context files (instructions, skills, agents, commands, settings, MCP) the commits changed; kinds and counts only (0.8.8) |
+| `quality.change_blocks`, `checked_blocks`, `rework_cycles`, `stage` | how many change blocks (edits with no check between them) a check followed, how often a failed check was followed by another change before it ran again (rework), and the furthest step the stretch reached here: attempted, verified, committed, pushed, pr, deployed; counts and one word only (0.8.9) |
 | `collector_version` | the CLI version that measured it |
 | `seq`, `prev`, `hash` | the line's place, the hash of the line before it, and its own hash |
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "worktrust",
-  "version": "0.8.8",
+  "version": "0.8.9",
   "description": "Couple this computer to WorkTrust: approve a code in your browser, and every AI app here reports what you built. Metadata only, no dependencies.",
   "type": "module",
   "bin": {

package/preserve-lines.mjs

@@ -105,7 +105,12 @@ export function archiveLine(entry, behaviour = null, { ids = {}, collector, salt
   // PRACTICE (0.8.8): framing, verification quality, risk, tool mix, re-reads, context files; whole counts and flags only.
   const bool = (value) => value === true || value === false;
   if (entry.framing && whole(entry.framing.turns_before_action) && bool(entry.framing.planned_first)) line.framing = { turns_before_action: entry.framing.turns_before_action, planned_first: entry.framing.planned_first };
-  if (entry.quality && whole(entry.quality.fix_cycles_max) && (bool(entry.quality.checked_after_change) || entry.quality.checked_after_change === null) && (bool(entry.quality.inspected_first) || entry.quality.inspected_first === null)) line.quality = { checked_after_change: entry.quality.checked_after_change, fix_cycles_max: entry.quality.fix_cycles_max, inspected_first: entry.quality.inspected_first };
+  if (entry.quality && whole(entry.quality.fix_cycles_max) && (bool(entry.quality.checked_after_change) || entry.quality.checked_after_change === null) && (bool(entry.quality.inspected_first) || entry.quality.inspected_first === null)) {
+    line.quality = { checked_after_change: entry.quality.checked_after_change, fix_cycles_max: entry.quality.fix_cycles_max, inspected_first: entry.quality.inspected_first };
+    // 0.8.9: change blocks, how many a check followed, rework cycles.
+    if (whole(entry.quality.change_blocks) && whole(entry.quality.checked_blocks) && whole(entry.quality.rework_cycles) && entry.quality.checked_blocks <= entry.quality.change_blocks) Object.assign(line.quality, { change_blocks: entry.quality.change_blocks, checked_blocks: entry.quality.checked_blocks, rework_cycles: entry.quality.rework_cycles });
+  }
+  if (STAGES.includes(entry.stage)) line.stage = entry.stage;
   const risk = pick(entry.risk, ["proposed", "refused", "run"]);
   if (risk && risk.proposed > 0 && risk.refused + risk.run <= risk.proposed) line.risk = risk;
   const keyed = (record, keys) => { const kept = Object.entries(record && typeof record === "object" ? record : {}).filter(([key, n]) => keys.includes(key) && whole(n) && n > 0).sort(); return kept.length ? Object.fromEntries(kept) : null; };
@@ -162,7 +167,9 @@ export function archiveEntries(dir, ownKey) {
  * a line of their own that names the stretch it adds to (`supplements`) and carries only what no earlier line of that
  * stretch carries, in the same chain, under the same day's root. Null when there is nothing new.
  */
-export const DERIVED_KEYS = ["signals", "analyzer_version", "verification", "unconfirmed", "delivery", "recovery", "delegation", "context", "routing", "steering", "tools", "complexity", "oversight", "planning", "changes", "autonomy", "authorship", "framing", "quality", "risk", "tool_mix", "reads", "context_files", "interrupts", "steers", "utc_offset"];
+/** The furthest step a stretch reached on the computer (0.8.9). */
+const STAGES = ["attempted", "verified", "committed", "pushed", "pr", "deployed"];
+export const DERIVED_KEYS = ["signals", "analyzer_version", "verification", "unconfirmed", "delivery", "recovery", "delegation", "context", "routing", "steering", "tools", "complexity", "oversight", "planning", "changes", "autonomy", "authorship", "framing", "quality", "risk", "tool_mix", "reads", "context_files", "stage", "interrupts", "steers", "utc_offset"];
 export function supplementFor(line, earlier) {
   const missing = DERIVED_KEYS.filter((key) => line[key] !== undefined && !earlier.some((old) => old[key] !== undefined));
   if (missing.length === 0) return null;

package/stretch-evidence.mjs

@@ -268,6 +268,10 @@ function autonomyOf(messages) {
  *            (plan mode, a plan approved, a to-do list) came before it.
  *   quality: whether a check passed AFTER the stretch's last change; the most failures one check family took before it
  *            passed; whether the change was looked at (git diff, show, status) before the first delivery step.
+ *            0.8.9: change BLOCKS (edits with no check between them) and how many a check followed before the next one;
+ *            REWORK cycles (a check failed, then a change before that check ran again: change, fail, change).
+ *   stage:   the furthest step the stretch reached on this computer: attempted (it changed something), verified (a check
+ *            passed), committed, pushed, pr, deployed (0.8.9). Merged and confirmed are the server's, from GitHub.
  *   risk:    destructive commands proposed, refused (by the person or a guard) and run.
  *   tool_mix: calls per kind of tool (search, read, edit, shell, agent, web, data, mcp, plan).
  *   reads:   files read again unchanged (the same read twice), the cost of context that did not hold.
@@ -282,6 +286,8 @@ function practiceOf(messages, { isHumanTurn }) {
   let turns = 0, acted = false, planned = false, turnsBefore = 0, plannedFirst = false;
   let lastChange = -1, passedAfter = false, delivered = false, inspectedFirst = false;
   const failuresBefore = new Map(); let cyclesMax = 0;
+  let blocks = 0, checkedBlocks = 0, inBlock = false, rework = 0, passedAny = false;
+  const pendingFail = new Set(), reached = new Set();
   const risk = { proposed: 0, refused: 0, run: 0 }, mix = {}, reads = new Map();
   let index = 0;
   for (const message of messages) {
@@ -294,15 +300,21 @@ function practiceOf(messages, { isHumanTurn }) {
       mix[toolKind(tool)] = (mix[toolKind(tool)] ?? 0) + 1;
       if (PLAN_TOOLS.has(tool) && tool !== "AskUserQuestion") planned = true;
       if (!acted && ACTION_TOOLS.has(tool)) { acted = true; turnsBefore = turns; plannedFirst = planned; }
-      if (EDIT_TOOLS.has(tool)) { lastChange = index; passedAfter = false; }
+      if (EDIT_TOOLS.has(tool)) {
+        lastChange = index; passedAfter = false;
+        if (!inBlock) { blocks += 1; inBlock = true; }
+        rework += pendingFail.size; pendingFail.clear();
+      }
       if (tool === "Read") reads.set(call.digest, (reads.get(call.digest) ?? 0) + 1);
       const checks = call.kinds.filter((kind) => CHECK_KINDS.includes(kind));
+      if (checks.length > 0 && inBlock) { checkedBlocks += 1; inBlock = false; }
       for (const kind of checks) {
-        if (result?.failed === true) failuresBefore.set(kind, (failuresBefore.get(kind) ?? 0) + 1);
+        if (result?.failed === true) { failuresBefore.set(kind, (failuresBefore.get(kind) ?? 0) + 1); pendingFail.add(kind); }
+        if (result?.failed === false) { passedAny = true; pendingFail.delete(kind); }
         if (result?.failed === false) { cyclesMax = Math.max(cyclesMax, failuresBefore.get(kind) ?? 0); failuresBefore.set(kind, 0); if (index > lastChange) passedAfter = true; }
       }
       if (call.kinds.includes("inspect") && !delivered) inspectedFirst = true;
-      if (call.kinds.some((kind) => DELIVERY_KINDS.includes(kind)) && result?.failed === false) delivered = true;
+      if (call.kinds.some((kind) => DELIVERY_KINDS.includes(kind)) && result?.failed === false) { delivered = true; for (const kind of call.kinds) if (DELIVERY_KINDS.includes(kind)) reached.add(kind); }
       if (call.kinds.includes("destructive")) {
         risk.proposed += 1;
         if (result?.refusal) risk.refused += 1; else if (result?.failed === false) risk.run += 1;
@@ -311,9 +323,11 @@ function practiceOf(messages, { isHumanTurn }) {
   }
   if (index === 0) return null;
   const repeated = [...reads.values()].reduce((sum, n) => sum + Math.max(0, n - 1), 0);
+  const stage = ["deploy", "pr", "push", "commit"].find((kind) => reached.has(kind)) ?? (lastChange > 0 ? (passedAny ? "verified" : "attempted") : null);
   return {
     framing: { turns_before_action: acted ? turnsBefore : turns, planned_first: acted ? plannedFirst : planned },
-    quality: { checked_after_change: lastChange > 0 ? passedAfter : null, fix_cycles_max: cyclesMax, inspected_first: delivered ? inspectedFirst : null },
+    quality: { checked_after_change: lastChange > 0 ? passedAfter : null, fix_cycles_max: cyclesMax, inspected_first: delivered ? inspectedFirst : null, change_blocks: blocks, checked_blocks: checkedBlocks, rework_cycles: rework },
+    ...(stage ? { stage: { deploy: "deployed", pr: "pr", push: "pushed", commit: "committed" }[stage] ?? stage } : {}),
     ...(risk.proposed > 0 ? { risk } : {}),
     tool_mix: mix,
     reads: { repeated },

package/worktrust.mjs

@@ -73,7 +73,7 @@ const command = args.find((arg, at) => !arg.startsWith("--") && !(at > 0 && VALU
 const flag = (name) => { const at = args.indexOf(`--${name}`); return at >= 0 ? args[at + 1] : undefined; };
 const has = (name) => args.includes(`--${name}`);
 /** This CLI's version, said to the door so the app can tell which computer runs an old one (check-cli-package holds it equal to package.json). */
-const CLI_VERSION = "0.8.8";
+const CLI_VERSION = "0.8.9";
 const ORIGIN = (flag("origin") ?? process.env.WORKTRUST_ORIGIN ?? "https://app.worktrust.io").replace(/\/$/, "");
 const MCP = flag("url") ?? process.env.WORKTRUST_MCP_URL ?? `${ORIGIN}/api/mcp`;
 const HOME_DIR = join(homedir(), ".worktrust");