From 4587c757b17d8a92df7c98ec37ace4c5500a4f5a Mon Sep 17 00:00:00 2001 From: Agando Date: Mon, 2 Nov 2026 21:21:24 +0100 Subject: fix(newos4): read_file pages were silently truncated; delegation no longer invents "completed" MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Two defects found by re-deriving a report that claimed both were still open. read_file paging (module-aiassist.js) - A page larger than the result clip was clipped AFTER read_file had computed next_offset. On the last page next_offset is null, so the model was told truncated:false while its tail had been removed: measured 11 890 of 19 999 chars delivered, returned:19999, no pointer to the rest - 8 109 chars lost with no way to ask for them. * the clip now sets truncated=true and next_offset=offset+keep when it removes the tail of a page that had no continuation (and never splits a surrogate pair) * read_file caps a page at the result budget (min(pageCap, toolResultChars-200)), so a page is always delivered whole and next_offset is exactly contiguous Live v4.6.143: 19 999-char file -> page 11 800 (next_offset 11 800), page 2 8 199 (next_offset null): 11 800 + 8 199 = 19 999, nothing withheld. A model asking for limit:32000 is clamped to 11 800 and told truncated:true. The ~500-char symptom was the 500-char floor: a caller passing limit:200 gets 500, which is why targeted sed -n reads looked better than the tool. Delegation reporting (module-orchestrator.js, module-aiassist.js) - chat() returning {content} was already fixed in 8a9333b, but three other places still substituted the word "completed" for an absent answer, so the symptom survived: delegateTask's outbox wrote status:'completed' even when the worker returned nothing; the delegate_task tool returned {success:true, result:"completed"}; orchestrate-parallel summarised an unmatched agent as {status:'ok', result:'completed (agent recycled)'} and a missing outbox as "completed (no outbox)". - Now: an empty/undefined answer is status 'failed' with error 'agent produced no answer' / 'no result in outbox' / 'agent not found — result unverifiable', and the tool returns success:false with result:null. Tests: test-delegation-reporting.js 6 (RED 3/6 against HEAD: D3/D4 success:true, D5 the literal "completed"), test-tool-context-bound 26 with E6-E8 (RED 24/2: truncated:false, next_offset null). maintain 27/27. --- modules/core/module-orchestrator.js | 35 ++++++++++++++++++++++++----------- 1 file changed, 24 insertions(+), 11 deletions(-) (limited to 'modules/core/module-orchestrator.js') diff --git a/modules/core/module-orchestrator.js b/modules/core/module-orchestrator.js index 9f23ceb..6b8ee18 100644 --- a/modules/core/module-orchestrator.js +++ b/modules/core/module-orchestrator.js @@ -1116,13 +1116,21 @@ class OrchestratorModule { if (r.status === 'fulfilled') { // Find the agent that handled this sub-task const agent = Object.values(this._agents).find(a => a.currentTask === sub.goal || (a.lastTask === sub.goal)); - if (agent) { - const outbox = this.terminal.fs.fsReadJSON(`/agents/${agent.id}/outbox.json`); - const snippet = outbox?.result ? outbox.result.substring(0, 200) : 'completed (no outbox)'; - summaries.push({ task: sub.goal, agent: agent.id, status: 'ok', result: snippet }); - } else { - summaries.push({ task: sub.goal, status: 'ok', result: 'completed (agent recycled)' }); + if (!agent) { + // "ok / completed (agent recycled)" told the caller a task succeeded + // when we could not even find its agent. + summaries.push({ task: sub.goal, status: 'failed', error: 'agent not found — result unverifiable' }); + continue; } + const outbox = this.terminal.fs.fsReadJSON(`/agents/${agent.id}/outbox.json`); + const body = outbox && typeof outbox.result === 'string' ? outbox.result : ''; + const snippet = body.substring(0, 200); + summaries.push({ + task: sub.goal, agent: agent.id, + status: body ? 'ok' : 'failed', + result: body || null, + error: body ? undefined : (outbox && outbox.error) || 'no result in outbox' + }); } else { summaries.push({ task: sub.goal, status: 'failed', error: r.reason?.message || 'unknown error' }); } @@ -1228,26 +1236,31 @@ class OrchestratorModule { const _agentTag = agent.id + '/' + (agent.role || 'worker'); const result = await aiModule.chat([`[DELEGATED] ${task}`], { agentTag: _agentTag }); try { aiModule._sessions.delete(_agentTag); } catch (_e) {} - agent.status = 'completed'; + // chat() resolves {content, toolTurns}. Anything that is not a non-empty answer + // is a failure — the old code substituted the literal "completed", so a worker + // that returned nothing was indistinguishable from one that finished. + const _answer = typeof result === 'string' ? result : ((result && result.content) || ''); + agent.status = _answer ? 'completed' : 'failed'; // Write result to outbox this.terminal.fs.writeFile(`agents/${agent.id}/outbox.json`, JSON.stringify({ - result: typeof result === 'string' ? result : (result?.content || 'completed'), - status: 'completed', + result: _answer || null, + status: _answer ? 'completed' : 'failed', + error: _answer ? null : 'agent produced no answer', timestamp: new Date().toISOString() })); // Record in memory graph const memgraph = this.terminal.moduleManager.modules['MemoryGraphModule']; if (memgraph) { - memgraph.recordTask(task, typeof result === 'string' ? result : (result?.content || ''), agent.id); + memgraph.recordTask(task, _answer, agent.id); memgraph.recordFileAccess(`/agents/${agent.id}/outbox.json`); } this.terminal.print(`✅ ${agent.id} completed task`, 'success'); // Return result for callers that await delegateTask - outboxResult = typeof result === 'string' ? result : (result?.content || 'completed'); + outboxResult = _answer; // '' when the worker returned nothing; callers see that // Check restart policy — if always, set back to idle for next task if (agent.restartPolicy === 'always') { -- cgit v1.2.3