@ai-dev-methodologies/rlp-desk 0.15.4 → 0.15.5
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/CHANGELOG.md +17 -1
- package/package.json +9 -3
- package/src/node/MANIFEST.txt +3 -0
- package/src/node/prompts/prompt-assembler.mjs +2 -2
- package/src/node/run.mjs +70 -3
- package/src/node/runner/campaign-main-loop.mjs +13 -2
- package/src/scripts/run_ralph_desk.zsh +5 -3
- package/docs/rlp-desk/internal/verification-policy-gap-analysis.md +0 -523
- package/docs/rlp-desk/internal/verification-strategy-research.md +0 -2097
- package/docs/rlp-desk/plans/cozy-gliding-trinket.md +0 -53
- package/docs/rlp-desk/plans/frolicking-churning-honey.md +0 -253
- package/docs/rlp-desk/plans/keen-sauteeing-snowflake.md +0 -245
- package/docs/rlp-desk/plans/mutable-booping-corbato.md +0 -163
- package/docs/rlp-desk/plans/rlp-desk-0.11-handoff-7fixes.md +0 -352
- package/docs/rlp-desk/plans/rlp-desk-0.11.1-tmux-pane-disappearance.md +0 -260
- package/docs/rlp-desk/plans/rlp-desk-elegant-papert-agent-a8cd695ffca2a3ad8.md +0 -84
- package/docs/rlp-desk/plans/rlp-desk-elegant-papert.md +0 -270
- package/docs/rlp-desk/plans/rlp-desk-tmux-flywheel-routing.md +0 -730
- package/docs/rlp-desk/plans/toasty-whistling-diffie-agent-a6814625642e956da.md +0 -201
- package/docs/rlp-desk/plans/toasty-whistling-diffie.md +0 -117
- package/docs/rlp-desk/plans/validated-snacking-crayon.md +0 -204
- package/examples/calculator/.claude/ralph-desk/logs/loop-test/iter-001.worker-output.log +0 -0
- package/examples/calculator/.claude/ralph-desk/logs/loop-test/iter-001.worker-prompt.md +0 -38
- package/examples/calculator/.claude/ralph-desk/logs/loop-test/iter-001.worker-trigger.sh +0 -28
- package/examples/calculator/.claude/ralph-desk/logs/loop-test/session-config.json +0 -25
- package/examples/calculator/.claude/ralph-desk/logs/loop-test/status.json +0 -10
- package/examples/calculator/.claude/ralph-desk/logs/loop-test/worker-heartbeat.json +0 -1
package/CHANGELOG.md
CHANGED
|
@@ -11,7 +11,23 @@ For pre-v0.15.4 versions, refer to `git log` and individual GitHub release notes
|
|
|
11
11
|
- Post-v0.15.6: remove `RLP_LIFECYCLE_METRICS` flag entirely (per plan v3 ADR follow-ups).
|
|
12
12
|
- Phase D.1 (handoff documents) + Phase D.2 (per-stage agent role specialization) — both deferred per `docs/plans/v0.15.4-release-runbook.md` §7.6.
|
|
13
13
|
|
|
14
|
-
## [0.15.
|
|
14
|
+
## [0.15.5] — 2026-06-17
|
|
15
|
+
|
|
16
|
+
Patch: fixes surfaced by a fresh-context live dogfood of the tmux and agent run modes, plus packaging hygiene.
|
|
17
|
+
|
|
18
|
+
### Fixed
|
|
19
|
+
- **Curl-install (`install.sh`) produced a Node leader that could not start.** `src/node/MANIFEST.txt` was missing three runtime modules, so a curl-installed `--mode agent` / Node leader threw `ERR_MODULE_NOT_FOUND` at startup. MANIFEST regenerated. (npm installs were unaffected.)
|
|
20
|
+
- **Tmux-mode leader could hang at startup.** A bare `> "$COST_LOG"` redirect runs zsh's `$NULLCMD` (`cat`), which blocks reading stdin when launched with an open TTY (interactive shell / tmux pane). Changed to `: > "$COST_LOG"`.
|
|
21
|
+
- **Worker sometimes printed the iteration signal instead of writing it.** The per-iteration prompt now explicitly instructs the worker to WRITE the verify signal to the iter-signal file (resolved path in tmux mode).
|
|
22
|
+
- **Clearer error for mis-leveled PRD user-story headings.** When user stories use `###` instead of `## US-NNN` (H2), the Node leader now hints at the correct heading level instead of a bare "No user stories found".
|
|
23
|
+
|
|
24
|
+
### Added
|
|
25
|
+
- **`clean <slug> [--kill-session]` for the Node CLI.** Resets a campaign — removes sentinels, signal/claim/verdict files, and runtime state, while preserving the PRD, test-spec, prompts, memory, and reports. Previously unimplemented in the Node rewrite, which left a blocked campaign with no recovery path.
|
|
26
|
+
|
|
27
|
+
### Changed
|
|
28
|
+
- **Smaller published package.** The tarball no longer ships internal research docs, dev planning handoffs, or example runtime state (73 → 53 files, ~364 → ~244 kB).
|
|
29
|
+
|
|
30
|
+
## [0.15.4] — 2026-05-08
|
|
15
31
|
|
|
16
32
|
Phase B: tmux/process lifecycle hardening + observability + real-LLM SV strengthening. 4 sequential PRs (B1, B2-FIX, B4, B3) merged to main, plus pre-release audit fix branch addressing 16 findings (3 CRITICAL, 6 HIGH, 5 MEDIUM, 2 LOW).
|
|
17
33
|
|
package/package.json
CHANGED
|
@@ -1,10 +1,12 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@ai-dev-methodologies/rlp-desk",
|
|
3
|
-
"version": "0.15.
|
|
3
|
+
"version": "0.15.5",
|
|
4
4
|
"description": "Fresh-context iterative loops for Claude Code — autonomous task completion with independent verification",
|
|
5
5
|
"scripts": {
|
|
6
6
|
"postinstall": "node scripts/postinstall.js",
|
|
7
7
|
"uninstall": "node scripts/uninstall.js",
|
|
8
|
+
"manifest:check": "node scripts/build-node-manifest.js --check",
|
|
9
|
+
"prepublishOnly": "node scripts/build-node-manifest.js --check",
|
|
8
10
|
"test:node": "node --test 'tests/node/*.mjs' 'tests/node/*.test.mjs'",
|
|
9
11
|
"test:zsh": "for f in tests/test_*.sh; do echo \"=== $f ===\"; zsh \"$f\" || exit 1; done",
|
|
10
12
|
"test:fast": "npm run test:node",
|
|
@@ -19,8 +21,12 @@
|
|
|
19
21
|
"src/governance.md",
|
|
20
22
|
"src/model-upgrade-table.md",
|
|
21
23
|
"scripts/",
|
|
22
|
-
"docs/rlp-desk
|
|
23
|
-
"
|
|
24
|
+
"docs/rlp-desk/*.md",
|
|
25
|
+
"docs/rlp-desk/blueprints/",
|
|
26
|
+
"examples/calculator/.claude/ralph-desk/context/",
|
|
27
|
+
"examples/calculator/.claude/ralph-desk/memos/",
|
|
28
|
+
"examples/calculator/.claude/ralph-desk/plans/",
|
|
29
|
+
"examples/calculator/.claude/ralph-desk/prompts/",
|
|
24
30
|
"install.sh",
|
|
25
31
|
"README.md",
|
|
26
32
|
"CHANGELOG.md",
|
package/src/node/MANIFEST.txt
CHANGED
|
@@ -7,9 +7,12 @@ reporting/campaign-reporting.mjs
|
|
|
7
7
|
run.mjs
|
|
8
8
|
runner/campaign-main-loop.mjs
|
|
9
9
|
runner/leader-registry.mjs
|
|
10
|
+
runner/prompt-detector.mjs
|
|
10
11
|
runner/prompt-dismisser.mjs
|
|
11
12
|
shared/fs.mjs
|
|
12
13
|
shared/paths.mjs
|
|
13
14
|
tmux/pane-manager.mjs
|
|
14
15
|
util/debug-log.mjs
|
|
16
|
+
util/desk-root.mjs
|
|
17
|
+
util/lifecycle-metrics.mjs
|
|
15
18
|
util/shell-quote.mjs
|
|
@@ -150,8 +150,8 @@ export async function assembleWorkerPrompt({
|
|
|
150
150
|
} else {
|
|
151
151
|
promptLines.push(`- **Test Spec**: Read \`${fullTestSpecPath}\` (full — find ${nextUs} section)`);
|
|
152
152
|
}
|
|
153
|
-
promptLines.push(`When done,
|
|
154
|
-
promptLines.push(`
|
|
153
|
+
promptLines.push(`When done, you MUST WRITE (not just print) the verify signal to the iter-signal FILE — Path: \`memos/<slug>-iter-signal.json\` (see the MANDATORY signal-file instruction in the base prompt).`);
|
|
154
|
+
promptLines.push(`Write this exact JSON to that file (us_id="${nextUs}", not "ALL"): {"iteration": N, "status": "verify", "us_id": "${nextUs}", "summary": "what was done", "timestamp": "ISO"}`);
|
|
155
155
|
promptLines.push('');
|
|
156
156
|
promptLines.push(`**Update the campaign memory's 'Next Iteration Contract' to reflect ${nextUs}.**`);
|
|
157
157
|
} else if (verifiedUs.length > 0) {
|
package/src/node/run.mjs
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
import fs from 'node:fs';
|
|
2
2
|
import os from 'node:os';
|
|
3
3
|
import path from 'node:path';
|
|
4
|
-
import { spawn } from 'node:child_process';
|
|
4
|
+
import { spawn, spawnSync } from 'node:child_process';
|
|
5
5
|
import { fileURLToPath } from 'node:url';
|
|
6
6
|
|
|
7
7
|
import { initCampaign } from './init/campaign-initializer.mjs';
|
|
@@ -9,6 +9,7 @@ import { readStatus } from './reporting/campaign-reporting.mjs';
|
|
|
9
9
|
import {
|
|
10
10
|
run as runCampaignMain,
|
|
11
11
|
detectLegacyDeskInRunMode,
|
|
12
|
+
buildPaths,
|
|
12
13
|
} from './runner/campaign-main-loop.mjs';
|
|
13
14
|
import { isClaudeEngine } from './cli/command-builder.mjs';
|
|
14
15
|
|
|
@@ -51,7 +52,7 @@ function buildHelpText() {
|
|
|
51
52
|
' run <slug> [options] Run loop (tmux=zsh leader [production], agent=Node leader [deprecated alpha], native=slash-only error)',
|
|
52
53
|
' status <slug> Show loop status',
|
|
53
54
|
' logs <slug> [N] Show iteration log (not implemented in the Node rewrite yet)',
|
|
54
|
-
' clean <slug> [--kill-session] Reset for re-run (
|
|
55
|
+
' clean <slug> [--kill-session] Reset for re-run (removes sentinels + runtime/; preserves PRD/prompts/memory)',
|
|
55
56
|
' resume <slug> Resume loop (not implemented in the Node rewrite yet)',
|
|
56
57
|
'',
|
|
57
58
|
'Run Options:',
|
|
@@ -217,6 +218,71 @@ async function runStatusCommand(args, deps) {
|
|
|
217
218
|
return 0;
|
|
218
219
|
}
|
|
219
220
|
|
|
221
|
+
// D-6 (dogfood): real `clean` for the Node leader. Previously "not implemented",
|
|
222
|
+
// which left a blocked campaign with NO recovery path (a transient parse error
|
|
223
|
+
// wrote a blocked sentinel that bricked re-runs). Removes the transient/terminal
|
|
224
|
+
// state (sentinels, signal/claim/verdict, runtime/) while PRESERVING the durable
|
|
225
|
+
// inputs (PRD, test-spec, prompts, context, memory) and the campaign report.
|
|
226
|
+
async function runCleanCommand(args, deps) {
|
|
227
|
+
if (args.length === 0 || args[0] === '--help') {
|
|
228
|
+
write(deps.stdout, 'Usage: node src/node/run.mjs clean <slug> [--kill-session]');
|
|
229
|
+
return 0;
|
|
230
|
+
}
|
|
231
|
+
const slug = args[0];
|
|
232
|
+
const killSession = args.includes('--kill-session');
|
|
233
|
+
const paths = buildPaths(deps.cwd, slug);
|
|
234
|
+
|
|
235
|
+
// --kill-session: read the session name from runtime/session-config.json
|
|
236
|
+
// BEFORE removing runtime, then best-effort tmux teardown.
|
|
237
|
+
if (killSession) {
|
|
238
|
+
try {
|
|
239
|
+
const cfgPath = path.join(paths.runtimeDir, 'session-config.json');
|
|
240
|
+
if (deps.fileExists(cfgPath)) {
|
|
241
|
+
const cfg = JSON.parse(fs.readFileSync(cfgPath, 'utf8'));
|
|
242
|
+
if (cfg && cfg.session_name) {
|
|
243
|
+
spawnSync('tmux', ['kill-session', '-t', cfg.session_name], { stdio: 'ignore' });
|
|
244
|
+
}
|
|
245
|
+
}
|
|
246
|
+
} catch { /* best-effort */ }
|
|
247
|
+
}
|
|
248
|
+
|
|
249
|
+
const transient = [
|
|
250
|
+
paths.blockedSentinel,
|
|
251
|
+
paths.blockedSentinel.replace(/\.md$/, '.json'),
|
|
252
|
+
paths.completeSentinel,
|
|
253
|
+
paths.completeSentinel.replace(/\.md$/, '.json'),
|
|
254
|
+
paths.signalFile,
|
|
255
|
+
paths.doneClaimFile,
|
|
256
|
+
paths.verdictFile,
|
|
257
|
+
paths.flywheelSignalFile,
|
|
258
|
+
paths.flywheelGuardVerdictFile,
|
|
259
|
+
];
|
|
260
|
+
let removed = 0;
|
|
261
|
+
for (const target of transient) {
|
|
262
|
+
try {
|
|
263
|
+
if (fs.existsSync(target)) {
|
|
264
|
+
// Sentinels may be chmod 0o444 (write-lock); relax before unlink.
|
|
265
|
+
try { fs.chmodSync(target, 0o644); } catch { /* ignore */ }
|
|
266
|
+
fs.rmSync(target, { force: true });
|
|
267
|
+
removed += 1;
|
|
268
|
+
}
|
|
269
|
+
} catch { /* best-effort per-file */ }
|
|
270
|
+
}
|
|
271
|
+
try {
|
|
272
|
+
if (fs.existsSync(paths.runtimeDir)) {
|
|
273
|
+
fs.rmSync(paths.runtimeDir, { recursive: true, force: true });
|
|
274
|
+
removed += 1;
|
|
275
|
+
}
|
|
276
|
+
} catch { /* best-effort */ }
|
|
277
|
+
|
|
278
|
+
write(
|
|
279
|
+
deps.stdout,
|
|
280
|
+
`Cleaned ${slug}: removed ${removed} transient artifact(s) (sentinels, signal/claim/verdict, runtime/`
|
|
281
|
+
+ `${killSession ? ', tmux session' : ''}). Preserved PRD, test-spec, prompts, context, memory, reports.`,
|
|
282
|
+
);
|
|
283
|
+
return 0;
|
|
284
|
+
}
|
|
285
|
+
|
|
220
286
|
// v0.14.0: Default location of the zsh runner installed by postinstall.js
|
|
221
287
|
// (Phase 3 of the v0.14.0 plan re-enables this sync). Overridable via
|
|
222
288
|
// RLP_DESK_ZSH_RUNNER for development checkouts that point to src/scripts.
|
|
@@ -466,9 +532,10 @@ export async function main(argv = process.argv.slice(2), overrides = {}) {
|
|
|
466
532
|
return await runRunCommand(rest, deps);
|
|
467
533
|
case 'status':
|
|
468
534
|
return await runStatusCommand(rest, deps);
|
|
535
|
+
case 'clean':
|
|
536
|
+
return await runCleanCommand(rest, deps);
|
|
469
537
|
case 'brainstorm':
|
|
470
538
|
case 'logs':
|
|
471
|
-
case 'clean':
|
|
472
539
|
case 'resume':
|
|
473
540
|
throw new Error(`${command} is not implemented in the Node rewrite yet`);
|
|
474
541
|
default:
|
|
@@ -93,7 +93,7 @@ export function detectLegacyDeskInRunMode(rootDir, env = process.env) {
|
|
|
93
93
|
return { legacyPath, newPath, message };
|
|
94
94
|
}
|
|
95
95
|
|
|
96
|
-
function buildPaths(rootDir, slug, env = process.env) {
|
|
96
|
+
export function buildPaths(rootDir, slug, env = process.env) {
|
|
97
97
|
const deskRoot = resolveDeskRoot(rootDir, env);
|
|
98
98
|
const campaignLogDir = path.join(deskRoot, 'logs', slug);
|
|
99
99
|
|
|
@@ -1462,7 +1462,18 @@ async function _runCampaignBody(slug, options, paths, rootDir) {
|
|
|
1462
1462
|
const usList = await readUsList(paths, slug);
|
|
1463
1463
|
|
|
1464
1464
|
if (usList.length === 0) {
|
|
1465
|
-
|
|
1465
|
+
// D-5 (dogfood): both leaders parse only H2 `## US-NNN:`. A common mistake is
|
|
1466
|
+
// authoring `### US-NNN` (H3+), which yields zero stories. Surface an actionable
|
|
1467
|
+
// hint instead of a bare "not found" (the zsh leader silently degrades here;
|
|
1468
|
+
// Node fail-closes — the safer behavior, now recoverable via `clean`).
|
|
1469
|
+
let hint = '';
|
|
1470
|
+
try {
|
|
1471
|
+
const prdRaw = await fs.readFile(paths.prdFile, 'utf8');
|
|
1472
|
+
if (/^#{3,}\s+US-\d{3}\b/m.test(prdRaw)) {
|
|
1473
|
+
hint = ' — found US-NNN heading(s) at level ### or deeper; US headings must be H2 ("## US-NNN:")';
|
|
1474
|
+
}
|
|
1475
|
+
} catch { /* best-effort hint */ }
|
|
1476
|
+
throw new Error(`No user stories found for ${slug}${hint}`);
|
|
1466
1477
|
}
|
|
1467
1478
|
|
|
1468
1479
|
if (!state.current_us) {
|
|
@@ -929,7 +929,9 @@ create_session() {
|
|
|
929
929
|
BASELINE_COMMIT=$(git -C "$ROOT" rev-parse HEAD 2>/dev/null || echo "none")
|
|
930
930
|
|
|
931
931
|
# Truncate cost-log for fresh run (previous data in versioned campaign reports)
|
|
932
|
-
>
|
|
932
|
+
# NOTE: ': >' not bare '>' — in zsh a bare redirect with no command runs $NULLCMD
|
|
933
|
+
# (=cat), which blocks reading stdin when the leader has an open TTY (D-1 dogfood hang).
|
|
934
|
+
: > "$COST_LOG"
|
|
933
935
|
|
|
934
936
|
# v5.7 §4.2: WITH_SELF_VERIFICATION=1 is hard-rejected at script entry now,
|
|
935
937
|
# so by the time we reach create_session() the flag is guaranteed to be 0.
|
|
@@ -1853,8 +1855,8 @@ write_worker_trigger() {
|
|
|
1853
1855
|
else
|
|
1854
1856
|
echo "- **Test Spec**: Read \`$DESK/plans/test-spec-${SLUG}.md\` (full — find ${next_us} section)"
|
|
1855
1857
|
fi
|
|
1856
|
-
echo "When done,
|
|
1857
|
-
echo "
|
|
1858
|
+
echo "When done, you MUST WRITE (not just print) the verify signal to the iter-signal FILE at: ${SIGNAL_FILE}"
|
|
1859
|
+
echo "Write this exact JSON to that file (us_id=\"${next_us}\", not \"ALL\"): {\"iteration\": N, \"status\": \"verify\", \"us_id\": \"${next_us}\", \"summary\": \"what was done\", \"timestamp\": \"ISO\"}"
|
|
1858
1860
|
echo ""
|
|
1859
1861
|
echo "**Update the campaign memory's 'Next Iteration Contract' to reflect ${next_us}.**"
|
|
1860
1862
|
elif [[ -n "$VERIFIED_US" ]]; then
|