diff --git a/bin/gstack-doctor-components.sh b/bin/gstack-doctor-components.sh
index 36caac274..d99e9c186 100644
--- a/bin/gstack-doctor-components.sh
+++ b/bin/gstack-doctor-components.sh
@@ -347,6 +347,7 @@ gstack_result_anchor() {
PREREG_INCOMPLETE) printf '%s' 'prereg-incomplete' ;;
PREREG_DIAGNOSTIC_UNTESTED) printf '%s' 'prereg-diagnostic-untested' ;;
PREREG_NO_GSTACK_ARM) printf '%s' 'prereg-no-gstack-arm' ;;
+ EVAL_REPORT_NO_GSTACK_ROW) printf '%s' 'eval-report-no-gstack-row' ;;
EVAL_ARM_DUPLICATE) printf '%s' 'eval-arm-duplicate' ;;
*) return 1 ;;
esac
diff --git a/bin/gstack-eval-plan b/bin/gstack-eval-plan
index 1dc081a76..020494a43 100755
--- a/bin/gstack-eval-plan
+++ b/bin/gstack-eval-plan
@@ -10,6 +10,7 @@
* gstack-eval-plan pilot settle --out
--attempt --usd
* gstack-eval-plan pilot release --out --attempt
* gstack-eval-plan pilot report --out --total [--workers W] [--json]
+ * gstack-eval-plan results --check [--json]
* gstack-eval-plan price-gate --gate-file | --gate "" [--plan-file ] [--base ] [--cap-usd N]
* [--smoke-cap-usd N] [--allow-unpriced ""] [--out ] [--json]
*
@@ -26,12 +27,13 @@
* price-gate lint every arm's model against the base branch's
* .gstack/eval-models.json (docs/eval-models.md; lib/eval-models.ts). Exit 0 ok, 1 fail (incomplete
* prereg, failed check, exceeded cap, unpriced or stale gate), 2 usage, 3 refused.
- * Logic: lib/eval-plan.ts.
+ * Logic: lib/eval-plan.ts; `results --check` in lib/eval-plan-report.ts.
*/
import * as fs from 'node:fs';
import * as path from 'node:path';
import { spawnSync } from 'node:child_process';
import { armModel, checkPrereg, DEFAULT_TOKENS, PREREG_SECTIONS, dryRun, pilotReport, preregTemplate, priceGate, releasePilot, renderArms, renderDryRun, renderGatePrice, renderPilotReport, renderPreregDiagnostics, resolveArms, resolvePlanSource, runPreregDiagnostics, settlePilot, startPilot } from '../lib/eval-plan';
+import { checkResults } from '../lib/eval-plan-report';
import { checkModelRules, EVAL_SCOPES, type EvalScope, type ModelRulesCheck } from '../lib/eval-models';
import { EXIT, exitTableHelp } from '../lib/headless-artifacts';
import { resultLine } from '../lib/result-codes';
@@ -41,7 +43,8 @@ import { SpendCapExceeded } from '../lib/spend-ledger';
const HELP = `gstack-eval-plan — preregistration, priced arms, $0 dry run, priced pilot (eval-plan/SKILL.md)
prereg --check [--allow-repo-commands] [--json]
- the five sections exist and are filled: Bars, Arms, Decoys, Stop rules, Held-out exposure;
+ the required sections exist and are filled: Bars, Arms, Comparison, Decoys, Stop rules,
+ Held-out exposure, and No-gstack arm when the Comparison says stack: loaded;
an optional Diagnostics section's summarizer runs on its synthetic fixture only with
--allow-repo-commands (else it is listed and exit 3) and every count must match
prereg --init [--title ] write the template
@@ -53,6 +56,9 @@ const HELP = `gstack-eval-plan — preregistration, priced arms, $0 dry run, pri
pilot start --out dir --cap usd --arms m,m#no-gstack --dry-run dry-run.json [--n 3]
pilot settle --out dir --attempt id --usd pilot release --out dir --attempt id
pilot report --out dir --total items [--workers W] [--json] spend table + extrapolation with a 95% band, then the approval ask
+ results --check [--json]
+ one table in the results file has a filled row for every Arms identity, each no-gstack arm
+ beside its variants; reads text only (docs/TESTING_INTERNALS.md#no-gstack-arm)
price-gate --gate-file | --gate "" [--plan-file f] [--base b] [--cap-usd N] [--smoke-cap-usd N] [--allow-unpriced ""] [--out dir] [--json]
price a paid gate from its GSTACK_EVAL_CALLS v1 call-plan file (--plan-file, else the base
policy's price_file at HEAD); checks the gate/head/input binding, executes nothing, writes a
@@ -129,6 +135,17 @@ function doPrereg(rest: string[]): number {
return EXIT.ok;
}
+function doResults(rest: string[]): number {
+ const prereg = flag(rest, '--check');
+ const results = rest.filter(a => !a.startsWith('--') && a !== prereg)[0];
+ if (!prereg || !results || !fs.existsSync(prereg) || !fs.existsSync(results)) return usage('results needs --check ');
+ const r = checkResults(fs.readFileSync(prereg, 'utf8'), fs.readFileSync(results, 'utf8'));
+ if (rest.includes('--json')) process.stdout.write(JSON.stringify({ prereg, results, ...r }) + '\n');
+ else process.stdout.write(`RESULTS: ${r.ok ? 'ok' : 'incomplete'} no_gstack_rows=${r.no_gstack.filter(id => r.found.includes(id)).length}/${r.no_gstack.length} rows=${r.found.length}/${r.required.length} file=${results}\n`);
+ if (!r.ok) { process.stderr.write(resultLine('gstack-eval-plan results', 'EVAL_REPORT_NO_GSTACK_ROW', r.problems.join('; ')) + '\n'); return EXIT.fail; }
+ return EXIT.ok;
+}
+
function doArms(rest: string[]): number {
const tokens = { input: num(rest, '--input-tokens', DEFAULT_TOKENS.input)!, output: num(rest, '--output-tokens', DEFAULT_TOKENS.output)! };
const scope = scopeOf(rest);
@@ -242,6 +259,7 @@ function main(argv: string[]): number {
if (cmd === 'dry-run') return doDryRun(rest);
if (cmd === 'pilot') return doPilot(rest);
if (cmd === 'price-gate') return doPriceGate(rest);
+ if (cmd === 'results') return doResults(rest);
return usage(`unknown command ${cmd}`);
} catch (e: any) {
if (e instanceof SpendCapExceeded) { process.stderr.write(resultLine(`gstack-eval-plan ${cmd}`, 'EVAL_BUDGET_EXHAUSTED', e.message) + '\n'); return EXIT.fail; }
diff --git a/docs/TESTING_INTERNALS.md b/docs/TESTING_INTERNALS.md
index 2ed948c98..bb2e1f2dc 100644
--- a/docs/TESTING_INTERNALS.md
+++ b/docs/TESTING_INTERNALS.md
@@ -1676,9 +1676,10 @@ own attempts. The arm gets:
**Reporting.** The no-gstack result is one row beside the variants in every
results table and every verdict, including when it wins. A comparison that
-drops the row has not reported its result. The prereg check cannot see the
-report, so the `report:` line states where the row goes and the reviewer of
-the result holds the author to it.
+drops the row has not reported its result. `gstack-eval-plan results --check
+ ` enforces it: one table in the results file must hold a
+filled row for every Arms identity (`EVAL_REPORT_NO_GSTACK_ROW` otherwise), and
+no number is published before it passes.
**Precedent and its gaps.** `test/helpers/arm-benchmark-harness.ts` already has
a `without-skill` arm beside `with-skill`, and it is the closest existing
diff --git a/docs/troubleshooting.md b/docs/troubleshooting.md
index 5f1fb6c28..f0a7ce5fe 100644
--- a/docs/troubleshooting.md
+++ b/docs/troubleshooting.md
@@ -3672,6 +3672,21 @@ goes beside the variants. When no arm loads a skill stack, write
`stack: none` with a `why:` line instead. The rule is in
[docs/TESTING_INTERNALS.md](TESTING_INTERNALS.md#no-gstack-arm).
+
+### `EVAL_REPORT_NO_GSTACK_ROW`
+
+**Meaning.** `gstack-eval-plan results --check ` found no
+single markdown table in the results file with a filled row for every arm
+identity the preregistration's Arms section lists. Usually the
+`#no-gstack` row is missing, sits in a separate table from its
+variants, or holds only placeholders (`-`, `n/a`, `tbd`). It also fires when
+the preregistration itself fails `prereg --check`.
+
+**Fix.** Put one row per arm identity, each no-gstack arm beside its variants,
+in one table, and rerun the check. Publish no number before it prints
+`RESULTS: ok`. A preregistration with `stack: none` needs no row. The rule is
+in [docs/TESTING_INTERNALS.md](TESTING_INTERNALS.md#no-gstack-arm).
+
### `EVAL_ARM_DUPLICATE`
diff --git a/eval-plan/SKILL.md b/eval-plan/SKILL.md
index 5ec40e511..66d7d8920 100644
--- a/eval-plan/SKILL.md
+++ b/eval-plan/SKILL.md
@@ -307,6 +307,20 @@ Under `SESSION_KIND: unattended` write the question as an `approval` gate item t
The full run, when approved, keeps settling every attempt into the same ledger and reprints
the spend table at each gate; `budget_exhausted` stops it.
+## Step 5: Results beside the no-gstack arm
+
+Write the run's results as one markdown table in a results file, one row per arm identity
+from the Arms section, each `#no-gstack` row beside its variants. Before any number
+leaves the run (a report, a READY block, a PR body), check it:
+
+```bash
+~/.claude/skills/gstack/bin/gstack-eval-plan results --check
+```
+
+`RESULTS: ok` lets the numbers go out. `EVAL_REPORT_NO_GSTACK_ROW` (exit 1) names the arm
+whose filled row is missing from the table that holds the others; add it and rerun. A
+preregistration with `stack: none` needs no row.
+
## Tier 3, after P7 (not in this skill yet)
Each of these waits for the adoption number in the multi-agent wave plan (P7); one line each so the shape is known:
diff --git a/eval-plan/SKILL.md.tmpl b/eval-plan/SKILL.md.tmpl
index 019499eca..8a2fc4d75 100644
--- a/eval-plan/SKILL.md.tmpl
+++ b/eval-plan/SKILL.md.tmpl
@@ -222,6 +222,20 @@ Under `SESSION_KIND: unattended` write the question as an `approval` gate item t
The full run, when approved, keeps settling every attempt into the same ledger and reprints
the spend table at each gate; `budget_exhausted` stops it.
+## Step 5: Results beside the no-gstack arm
+
+Write the run's results as one markdown table in a results file, one row per arm identity
+from the Arms section, each `#no-gstack` row beside its variants. Before any number
+leaves the run (a report, a READY block, a PR body), check it:
+
+```bash
+~/.claude/skills/gstack/bin/gstack-eval-plan results --check
+```
+
+`RESULTS: ok` lets the numbers go out. `EVAL_REPORT_NO_GSTACK_ROW` (exit 1) names the arm
+whose filled row is missing from the table that holds the others; add it and rerun. A
+preregistration with `stack: none` needs no row.
+
## Tier 3, after P7 (not in this skill yet)
Each of these waits for the adoption number in the multi-agent wave plan (P7); one line each so the shape is known:
diff --git a/lib/eval-plan-report.ts b/lib/eval-plan-report.ts
new file mode 100644
index 000000000..fef62483f
--- /dev/null
+++ b/lib/eval-plan-report.ts
@@ -0,0 +1,46 @@
+/**
+ * eval-plan-report: the results check behind `gstack-eval-plan results --check`.
+ * A preregistration that says `stack: loaded` names its no-gstack arms; its
+ * results file must hold one markdown table with a filled row for every Arms
+ * identity, so each no-gstack row sits beside its variants
+ * (docs/TESTING_INTERNALS.md#no-gstack-arm). Reads text only; executes nothing.
+ */
+import { armIdentities, checkPrereg } from './eval-plan';
+
+export interface ResultsCheck { ok: boolean; required: string[]; no_gstack: string[]; found: string[]; missing: string[]; problems: string[] }
+
+const PLACEHOLDER = /^(?:|-+|—|–|n\/a|tbd|todo|\?)$/i;
+const cellsOf = (line: string) => line.trim().replace(/^\|/, '').replace(/\|$/, '').split('|').map(c => c.trim());
+const separator = (line: string | undefined) => !!line && /^\s*\|[\s:|-]+\|?\s*$/.test(line) && line.includes('-');
+
+/** The filled-row identities of each markdown table: first cell, backticks stripped, header and separator rows skipped. */
+export function resultTables(text: string): Array> {
+ const lines = text.split('\n');
+ const tables: Array> = [];
+ let current: Set | null = null;
+ lines.forEach((line, i) => {
+ if (!/^\s*\|/.test(line)) { current = null; return; }
+ if (!current) { current = new Set(); tables.push(current); }
+ if (separator(line) || separator(lines[i + 1])) return;
+ const [first, ...rest] = cellsOf(line);
+ const id = (first ?? '').replace(/`/g, '').trim();
+ if (id && rest.some(c => !PLACEHOLDER.test(c.replace(/`/g, '').trim()))) current.add(id);
+ });
+ return tables;
+}
+
+export function checkResults(preregText: string, resultsText: string): ResultsCheck {
+ const prereg = checkPrereg(preregText);
+ if (prereg.comparison.stack === 'none') return { ok: true, required: [], no_gstack: [], found: [], missing: [], problems: [] };
+ const required = [...new Set(armIdentities(preregText))];
+ const noGstack = prereg.no_gstack.arms;
+ const problems: string[] = [];
+ if (!prereg.ok) problems.push('the preregistration does not pass prereg --check');
+ const tables = resultTables(resultsText);
+ const best = tables.reduce>((b, t) => (required.filter(id => t.has(id)).length > required.filter(id => b.has(id)).length ? t : b), new Set());
+ const found = required.filter(id => best.has(id));
+ const missing = required.filter(id => !best.has(id));
+ if (!tables.length) problems.push('no markdown table in the results file');
+ else for (const id of missing) problems.push(`no filled row for ${id} in the table that holds the other arms`);
+ return { ok: problems.length === 0, required, no_gstack: noGstack, found, missing, problems };
+}
diff --git a/lib/result-codes-eval.ts b/lib/result-codes-eval.ts
index 70102382b..3dac8edd2 100644
--- a/lib/result-codes-eval.ts
+++ b/lib/result-codes-eval.ts
@@ -25,6 +25,11 @@ export const EVAL_RESULT_CODES = {
summary: 'the Comparison says stack: loaded and the No-gstack arm section is missing, lacks a key, or does not give every model its own #no-gstack Arms line',
fix: 'add `## No-gstack arm` with arms:, inputs:, prompt:, panel: and report:, and list each #no-gstack arm under Arms; when no arm loads a skill stack, say stack: none with a why: line',
},
+ EVAL_REPORT_NO_GSTACK_ROW: {
+ anchor: 'eval-report-no-gstack-row',
+ summary: 'the results file has no single table with a filled row for every arm the preregistration lists, so a no-gstack row is missing or not beside its variants',
+ fix: 'put one row per arm identity (each #no-gstack beside its variants) in one markdown table of the results file, then rerun `gstack-eval-plan results --check`; publish no number before it says ok',
+ },
EVAL_ARM_DUPLICATE: {
anchor: 'eval-arm-duplicate',
summary: 'the pilot was given the same arm identity twice, so two arms would share one set of attempts',
diff --git a/test/eval-plan-results.test.ts b/test/eval-plan-results.test.ts
new file mode 100644
index 000000000..1544da203
--- /dev/null
+++ b/test/eval-plan-results.test.ts
@@ -0,0 +1,96 @@
+/**
+ * GST-471: `gstack-eval-plan results --check`. A loaded preregistration's
+ * results file must hold one markdown table with a filled row for every Arms
+ * identity, so each no-gstack row sits beside its variants.
+ */
+import { afterEach, describe, expect, test } from 'bun:test';
+import { spawnSync } from 'node:child_process';
+import * as fs from 'node:fs';
+import * as os from 'node:os';
+import * as path from 'node:path';
+import { checkResults } from '../lib/eval-plan-report';
+
+const BIN = path.resolve(import.meta.dir, '..', 'bin', 'gstack-eval-plan');
+const dirs: string[] = [];
+afterEach(() => { while (dirs.length) fs.rmSync(dirs.pop()!, { recursive: true, force: true }); });
+
+const BASE = ['Bars', 'Decoys', 'Stop rules', 'Held-out exposure'].map(s => `## ${s}\n- filled in\n`).join('\n');
+const PREREG = BASE + `
+## Arms
+- claude-opus-5-5#review-g default settings
+- claude-opus-5-5#no-gstack default settings
+
+## Comparison
+stack: loaded
+
+## No-gstack arm
+arms: claude-opus-5-5#no-gstack
+inputs: the same rows
+prompt: the task only
+panel: rule, 3 trials, same judge
+report: results.md, one row beside the variant
+`;
+const NONE = BASE + '\n## Arms\n- claude-opus-5-5 default\n\n## Comparison\nstack: none\nwhy: scorer calibration\n';
+const table = (...rows: string[]) => ['| arm | passed | cost |', '|---|---|---|', ...rows].join('\n') + '\n';
+
+describe('checkResults', () => {
+ test('one table with a filled row per arm passes; backticks and the header row are handled', () => {
+ const r = checkResults(PREREG, '# Results\n\n' + table('| `claude-opus-5-5#review-g` | 7/9 | $1.20 |', '| claude-opus-5-5#no-gstack | 5/9 | $0.80 |'));
+ expect(r).toMatchObject({ ok: true, no_gstack: ['claude-opus-5-5#no-gstack'], missing: [] });
+ });
+
+ test('a missing no-gstack row fails naming it', () => {
+ const r = checkResults(PREREG, table('| claude-opus-5-5#review-g | 7/9 | $1.20 |'));
+ expect(r.ok).toBe(false);
+ expect(r.problems).toEqual(['no filled row for claude-opus-5-5#no-gstack in the table that holds the other arms']);
+ });
+
+ test('placeholder cells do not count as a filled row', () => {
+ for (const cell of ['', '-', '—', 'n/a', 'TBD']) {
+ const r = checkResults(PREREG, table('| claude-opus-5-5#review-g | 7/9 | $1.20 |', `| claude-opus-5-5#no-gstack | ${cell} | ${cell} |`));
+ expect(r.missing, JSON.stringify(cell)).toEqual(['claude-opus-5-5#no-gstack']);
+ }
+ });
+
+ test('the no-gstack row must sit in the same table as its variants; prose mentions never count', () => {
+ const split = table('| claude-opus-5-5#review-g | 7/9 | $1.20 |') + '\nThe claude-opus-5-5#no-gstack arm passed 5/9.\n\n' + table('| claude-opus-5-5#no-gstack | 5/9 | $0.80 |');
+ expect(checkResults(PREREG, split).ok).toBe(false);
+ expect(checkResults(PREREG, 'claude-opus-5-5#review-g 7/9, claude-opus-5-5#no-gstack 5/9\n').problems).toEqual(['no markdown table in the results file']);
+ });
+
+ test('identities match whole: a prefix model does not stand in for the arm', () => {
+ const r = checkResults(PREREG, table('| claude-opus-5-5#review-g | 7/9 | $1.20 |', '| claude-opus-5#no-gstack | 5/9 | $0.80 |'));
+ expect(r.missing).toEqual(['claude-opus-5-5#no-gstack']);
+ });
+
+ test('stack: none needs no row; a preregistration that fails prereg --check fails here too', () => {
+ expect(checkResults(NONE, 'no table at all\n').ok).toBe(true);
+ const broken = PREREG.replace(/## No-gstack arm[\s\S]*$/, '');
+ const r = checkResults(broken, table('| claude-opus-5-5#review-g | 7/9 | $1.20 |', '| claude-opus-5-5#no-gstack | 5/9 | $0.80 |'));
+ expect(r.problems).toEqual(['the preregistration does not pass prereg --check']);
+ });
+});
+
+describe('gstack-eval-plan results --check', () => {
+ const run = (prereg: string, results: string) => {
+ const d = fs.mkdtempSync(path.join(os.tmpdir(), 'gstack-results-check-'));
+ dirs.push(d);
+ fs.writeFileSync(path.join(d, 'prereg.md'), prereg);
+ fs.writeFileSync(path.join(d, 'results.md'), results);
+ const r = spawnSync(BIN, ['results', '--check', 'prereg.md', 'results.md'], { cwd: d, encoding: 'utf8', timeout: 60_000 });
+ return { code: r.status, out: r.stdout ?? '', err: r.stderr ?? '' };
+ };
+
+ test('missing row: RESULTS: incomplete, EVAL_REPORT_NO_GSTACK_ROW, exit 1', () => {
+ const r = run(PREREG, table('| claude-opus-5-5#review-g | 7/9 | $1.20 |'));
+ expect(r.code).toBe(1);
+ expect(r.out).toBe('RESULTS: incomplete no_gstack_rows=0/1 rows=1/2 file=results.md\n');
+ expect(r.err).toContain('(EVAL_REPORT_NO_GSTACK_ROW)');
+ });
+
+ test('complete: RESULTS: ok, exit 0', () => {
+ const r = run(PREREG, table('| claude-opus-5-5#review-g | 7/9 | $1.20 |', '| claude-opus-5-5#no-gstack | 5/9 | $0.80 |'));
+ expect(r.code, r.err).toBe(0);
+ expect(r.out).toBe('RESULTS: ok no_gstack_rows=1/1 rows=2/2 file=results.md\n');
+ });
+});
diff --git a/test/fixtures/module-size-ratchet.json b/test/fixtures/module-size-ratchet.json
index d56450722..8139bdb26 100644
--- a/test/fixtures/module-size-ratchet.json
+++ b/test/fixtures/module-size-ratchet.json
@@ -34,6 +34,7 @@
{ "file": "lib/design-lint/rules/index.ts", "movedFrom": "(new: the bundled design lint, GSTA-105; ported from hallmark sloplint.mjs)" },
{ "file": "lib/design-lint/rules/copy.ts", "movedFrom": "(new: the bundled design lint, GSTA-105; ported from hallmark sloplint.mjs)" },
{ "file": "lib/pricing.ts", "movedFrom": "test/helpers/pricing.ts (PRICING, estimateCostUsd; plan E1 moved the table under lib/ so bin/gstack-eval-plan can price arms)" },
+ { "file": "lib/eval-plan-report.ts", "movedFrom": "(new: the results check behind gstack-eval-plan results --check, the no-gstack row beside its variants; GST-471)" },
{ "file": "lib/eval-plan.ts", "movedFrom": "(new: preregistration check, priced arms, $0 dry run, priced pilot ledger behind bin/gstack-eval-plan, plan E1)" },
{ "file": "lib/lane-ownership.ts", "movedFrom": "(new: lane ownership check behind bin/gstack-lane-check, plan E2)" },
{ "file": "lib/coordinator-contract.ts", "movedFrom": "(new: the inherited coordinator contract text, plan E2)" },