great-cto 3.74.0 → 3.75.0

package/board/.claude-plugin/plugin.json

@@ -1,7 +1,7 @@
 {
   "name": "great-cto",
   "description": "You already have the agent. This is everything around it. great_cto runs Claude Code as a pipeline of 72 specialist agents \u2014 an independent model checks each stage before the next builds on it, spending caps refuse rather than warn, and three decisions stay yours: what gets built, how, and whether it ships.",
-  "version": "3.74.0",
+  "version": "3.75.0",
   "author": {
     "name": "Alexander Velikiy",
     "url": "https://hashnode.com/@Greatcto"

package/board/agents/code-reviewer.md

@@ -92,8 +92,9 @@ snapshots, binaries) are excluded, because their size is not review effort.
 3. **Performance** — needless O(n²), N+1 queries, unbounded growth, sync work on a
    hot path, missing pagination/limits. Only when it matters at realistic scale.
 4. **Readability / maintainability** — naming, dead code, duplicated logic,
-   missing tests for new behaviour, comments that lie, an abstraction that hides a
-   bug. Match the surrounding code's idiom.
+   missing tests for new behaviour, a test an empty body would satisfy (it still
+   passes if the function under test does nothing, so it asserts nothing), comments
+   that lie, an abstraction that hides a bug. Match the surrounding code's idiom.
 
 ## Calibration — apply `shared/agent-fragments/argument-quality.md`
 

package/board/agents/project-auditor.md

@@ -227,7 +227,9 @@ Runs all of Phase 4 (sections labeled in stdout):
 - **4A.0** intersects the top-20 largest source files with the top-20 most-git-churned files
   (last 6 months), writing scratch files to `/tmp/audit-{largest,churn,hotspots}.txt`. Every
   file in the intersection MUST receive concrete file:line citations in the findings table —
-  these are the most expensive places to leave debt unaddressed.
+  these are the most expensive places to leave debt unaddressed. A hotspot with no test
+  beside it gets "characterize before change" in its Recommendation: the first edit there
+  should be a test pinning its current behaviour.
 - **4A** code structure signals — god files (>500 lines), circular-import risk count,
   dead-code marker count (TODO/FIXME/HACK/XXX/DEPRECATED), duplicated-basename files.
 - **4B** infrastructure & ops debt — Docker base image lines, k8s/Helm apiVersion usage,

package/board/agents/qa-engineer.md

@@ -611,6 +611,11 @@ asserts nothing. Mutation testing closes that — it introduces small faults
 (flip a comparison, drop a call, change a boundary) and reports how many the
 suite **catches**. A surviving mutant is a fault your tests would have shipped.
 
+Small tasks skip the run, not the question: the cheap form is the empty-implementation
+check — stub the function under test to do nothing; a test that still passes asserts
+nothing. senior-dev runs it at RED and code-reviewer flags it; if you see such a test
+in a small task, report it.
+
 Run it on the **changed files only** — a full-repo mutation run is minutes-to-hours
 and mostly re-verifies untouched code:
 

package/board/agents/senior-dev.md

@@ -73,7 +73,8 @@ that get pushed back on, because those are the ones nobody else will catch.
   projects measured on 2026-09-21, 1,181 of 5,126 operator messages were "делай" /
   "да" answering exactly that question.
 - the next step of a plan that already passed `gate:plan`;
-- re-running a check after a fix, until it is green or the failure is understood.
+- re-running a check after a fix, until it is green or the failure is understood (two failed
+  fixes on one assumption end that loop — see "Two fixes, one premise", step 5).
 
 This list does not replace **Ask before proceeding** above: a lint rule to disable, a
 pattern to copy, an "unrelated" failing test, a field added to an API, an empty
@@ -475,6 +476,11 @@ Schema: `skills/great_cto/references/knowledge-extraction.md`
    - If `data-platform` archetype → dbt tests + data contract validation / snapshot regression
    - If `ai-system` archetype → evals-first (write eval suite before any prompt/model changes)
    - All other archetypes → standard TDD (follow `superpowers:test-driven-development`):
+     - **Characterize first** — if the code you must change has no test covering it, write
+       a test that pins what it does TODAY (a golden/snapshot of current output is enough; it
+       records behaviour, not a verdict on it) and see it **pass on the unchanged code**. This
+       is the one test meant to pass before the change. Then change under RED/GREEN; when the
+       change alters the pinned output, update the pin deliberately and say so.
      - **RED**: write failing test, then **explicitly verify it fails**:
        ```bash
        npm test -- --testPathPattern="<new-test>" 2>&1 | grep -E "FAIL|PASS|Error" | head -5
@@ -488,9 +494,23 @@ Schema: `skills/great_cto/references/knowledge-extraction.md`
        PYTHONUNBUFFERED=1 pytest ... > /tmp/test-out.log 2>&1 &
        sleep 2; tail -f /tmp/test-out.log   # then Ctrl-C when done
        ```
-       If test PASSES before implementation → the test is wrong. Fix it first.
+       If test PASSES before implementation → the test is wrong. Fix it first. (The one
+       exception is the characterization pin above, which must pass on unchanged code.)
+     - **Stub check — before GREEN, even when RED is already confirmed.** A RED that failed
+       because the function does not exist ("not defined", an import error) says nothing about
+       the assertions. So create the stub (empty body: returns nothing, does nothing) and run
+       the new tests against it, then say in your reply which of them still pass. "Not
+       called", "does not throw" and "is defined" assertions all pass against it; a suite made
+       of them asserts nothing, so add the test for the positive effect (the call made, the
+       value returned) until the stub fails on an assertion. Mutation testing is skipped for
+       small tasks; this check is not.
      - **GREEN**: write minimal code to make test pass → re-run → confirm PASS
      - **REFACTOR**: clean up → confirm still PASS, no coverage regression
+     - **Two fixes, one premise → stop.** When two fixes resting on the same assumption have
+       failed the same check, write that assumption in one sentence and make no third edit on
+       it. Test the assumption itself first (print the value, query the real dependency, run
+       the one-line repro); if it is false, redesign from what you found, and if you cannot
+       test it, escalate. A third try on the same idea is a coin toss, not a fix.
 6. **Quality check**: all tests pass, no lint errors, coverage ≥80%, no hardcoded secrets
 6b. **Scope check against the IMPL-BRIEF** (governance Phase 3 — before staging):
    ```bash

package/board/scripts/lib/receipt.mjs

@@ -34,7 +34,17 @@ import { fileURLToPath } from 'node:url';
 
 const sha = (s) => createHash('sha256').update(String(s)).digest('hex');
 
+// Hooks read evidence under host deadlines of 5-15 s and must give up in time to
+// fail closed themselves, so 5 s is the default. A long-lived caller on a loaded
+// machine (the packaged controller probe) widens it with
+// GREAT_CTO_GIT_READ_TIMEOUT_MS; it never narrows below 5 s nor passes 120 s.
+export const GIT_READ_TIMEOUT_MS = 5000;
+export function gitReadTimeoutMs(raw = process.env.GREAT_CTO_GIT_READ_TIMEOUT_MS) {
+  return /^\d{1,9}$/.test(raw ?? '') ? Math.min(Math.max(Number(raw), GIT_READ_TIMEOUT_MS), 120000) : GIT_READ_TIMEOUT_MS;
+}
+
 function git(args, cwd, { maxBuffer = 32 * 1024 * 1024 } = {}) {
+  const timeout = gitReadTimeoutMs();
   try {
     // Reading evidence must not execute project-configured diff/filter/monitor helpers.
     const command = args[0] === 'diff' ? ['diff', '--no-ext-diff', '--no-textconv', ...args.slice(1)] : args;
@@ -45,14 +55,14 @@ function git(args, cwd, { maxBuffer = 32 * 1024 * 1024 } = {}) {
       let keys = '';
       try {
         keys = execFileSync('git', [...config, 'config', '--null', '--name-only', '--get-regexp', '^filter\\..*\\.(clean|process|required)$'],
-          { cwd, encoding: 'utf8', maxBuffer: 65536, timeout: 5000, stdio: ['ignore', 'pipe', 'ignore'] });
+          { cwd, encoding: 'utf8', maxBuffer: 65536, timeout, stdio: ['ignore', 'pipe', 'ignore'] });
       } catch (error) { if (error.status !== 1) return null; }
       const filters = [...new Set(keys.split('\0').filter(Boolean).map(key => key.match(/^filter\.([\s\S]+)\.(?:clean|process|required)$/)?.[1]))];
       if (filters.length > 128 || filters.some(name => !name)) return null;
       for (const name of filters) config.push('-c', `filter.${name}.clean=`, '-c', `filter.${name}.process=`, '-c', `filter.${name}.required=false`);
     }
     return execFileSync('git', [...config, ...command],
-      { cwd, encoding: 'utf8', maxBuffer, timeout: 5000, stdio: ['ignore', 'pipe', 'ignore'] });
+      { cwd, encoding: 'utf8', maxBuffer, timeout, stdio: ['ignore', 'pipe', 'ignore'] });
   } catch { return null; }
 }
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "great-cto",
-  "version": "3.74.0",
+  "version": "3.75.0",
   "description": "One command install for the great_cto Claude Code plugin. Auto-detects your stack, picks the right archetype, bootstraps PROJECT.md.",
   "keywords": [
     "claude-code",