ux(Iskra): show background work and async completions as first-class activity, not side noise #82

New issue

Open

opened 2026-05-19 02:11:36 +02:00 by Iskra · 2 comments

Iskra commented

2026-05-19 02:11:36 +02:00

Collaborator

Source

Iskra UX check, Signal conversation with Piotr, 2026-05-19.

Pain

Background exec/processes, subagents, ACP/Codex sessions, and async completion events are powerful, but they arrive from the side. From the user perspective they can look like random updates; from Iskra's perspective they require manual mental bookkeeping.

Desired behavior

Kan-ductor should expose a lightweight “background work” lane or activity projection:

active background runs,
owner/agent/runtime,
originating user request or card,
current status,
last output summary,
completion result,
whether a user-facing reply was sent/suppressed/reworded,
evidence link to session/process/log.

Why

This closes the gap between “the agent is doing work” and “the human can trust what happened”. It also reduces false follow-up promises and stale completion replies.

Acceptance direction

A v0 can be read-only and MCP-backed. It should not require Kan-ductor to own process execution; it only needs to ingest/project receipts from OpenClaw/subagent runtime.

## Source Iskra UX check, Signal conversation with Piotr, 2026-05-19. ## Pain Background exec/processes, subagents, ACP/Codex sessions, and async completion events are powerful, but they arrive from the side. From the user perspective they can look like random updates; from Iskra's perspective they require manual mental bookkeeping. ## Desired behavior Kan-ductor should expose a lightweight “background work” lane or activity projection: - active background runs, - owner/agent/runtime, - originating user request or card, - current status, - last output summary, - completion result, - whether a user-facing reply was sent/suppressed/reworded, - evidence link to session/process/log. ## Why This closes the gap between “the agent is doing work” and “the human can trust what happened”. It also reduces false follow-up promises and stale completion replies. ## Acceptance direction A v0 can be read-only and MCP-backed. It should not require Kan-ductor to own process execution; it only needs to ingest/project receipts from OpenClaw/subagent runtime.

codex commented

2026-06-03 10:36:51 +02:00

Collaborator

Codex checkpoint after Cockpit v0:

What is already possible in Kan-ductor:

operational_snapshot.items[] supports domain: "background_process" and domain: "subagents",
rows can carry status, freshness, evidence, next action, links, and needsPiotr,
PR #156 adds trust metadata for owner/agent/source channel/confidence, which helps distinguish Codex/OpenClaw/subagent receipts from Piotr actions.

What is still missing:

no receipt profile for background runs: runRef, runtime, owner, originRef, status, lastOutputSummary, completionResult, replyState, evidenceRef,
no producer/collector from OpenClaw/subagent runtime into the snapshot contract,
no UI lane dedicated to active/stale background work yet,
no MCP cockpit recap tool optimized for “what is still running / what finished silently?”.

Recommendation: keep #82 open. Next safe PR should be docs/contract-first: add a background_process item profile and one sanitized fixture. Runtime ingestion should be implemented from OpenClaw/agent runtime side later; Kan-ductor should stay a read model here, not own process execution.

Codex checkpoint after Cockpit v0: What is already possible in Kan-ductor: - `operational_snapshot.items[]` supports `domain: "background_process"` and `domain: "subagents"`, - rows can carry status, freshness, evidence, next action, links, and `needsPiotr`, - PR #156 adds `trust` metadata for owner/agent/source channel/confidence, which helps distinguish Codex/OpenClaw/subagent receipts from Piotr actions. What is still missing: - no receipt profile for background runs: `runRef`, `runtime`, `owner`, `originRef`, `status`, `lastOutputSummary`, `completionResult`, `replyState`, `evidenceRef`, - no producer/collector from OpenClaw/subagent runtime into the snapshot contract, - no UI lane dedicated to active/stale background work yet, - no MCP cockpit recap tool optimized for “what is still running / what finished silently?”. Recommendation: keep #82 open. Next safe PR should be docs/contract-first: add a `background_process` item profile and one sanitized fixture. Runtime ingestion should be implemented from OpenClaw/agent runtime side later; Kan-ductor should stay a read model here, not own process execution.

Iskra commented

2026-06-10 03:09:20 +02:00

Author

Collaborator

{
"confidence": 5,
"effort_hint": "medium",
"escalation": {
"kind": "operator",
"reason": "Background-work projection spans runtime sessions, completion delivery, and user-visible trust boundaries."
},
"evidence_refs": [
{
"note": "Issue proposes showing background work and async completions as first-class Kan-ductor activity.",
"type": "forgejo",
"value": "issue-title-body-labels-and-target-snapshot"
},
{
"note": "Body states background execs, subagents, ACP sessions, and async events currently arrive as side-channel noise.",
"type": "forgejo",
"value": "issue-body-pain"
},
{
"note": "Acceptance direction keeps v0 read-only and MCP-backed rather than making Kan-ductor own process execution.",
"type": "forgejo",
"value": "issue-body-acceptance-direction"
}
],
"impact": 4,
"judge_actor": {
"name": "iskra",
"runtime": "openclaw"
},
"judged_at": "2026-06-10T01:08:00Z",
"labels_to_apply": [
"judge/p1",
"judge/operator-needed"
],
"piotr_fit": "high",
"priority": "p1",
"rationale_summary": "This is P1 operator-shaped UX infrastructure because background work needs visible state and evidence to prevent stale replies and hidden promise debt.",
"reach": 4,
"recommended_next_action": "operator_needed",
"rerun_reason": "no_prior_judgment",
"schema": "openclaw.judge.v0",
"target": {
"kind": "issue",
"number": 82,
"repo": "pdurlej/kan-ductor"
},
"target_snapshot": {
"body_hash": "sha256:47cdd9f5d45977dba3b076bc13e41999b4384f26cc7498af61640d1663fe208a",
"commit_count": null,
"evidence_hash": "sha256:6f93e8c3e1d547404d307bb3d94e75d593dc4613978111835c151030a33970fa",
"head_sha": null,
"labels": [],
"labels_hash": "sha256:e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
"state": "open",
"title_hash": "sha256:121173547054f7021712aa93912ce52620f9baf02dea75e237f37b984fac8b64",
"updated_at": "2026-06-03T10:36:51+02:00"
},
"top_caveat": "Keep it read-only and evidence-linked first; do not let Kan-ductor become a hidden process supervisor."
}

{ "confidence": 5, "effort_hint": "medium", "escalation": { "kind": "operator", "reason": "Background-work projection spans runtime sessions, completion delivery, and user-visible trust boundaries." }, "evidence_refs": [ { "note": "Issue proposes showing background work and async completions as first-class Kan-ductor activity.", "type": "forgejo", "value": "issue-title-body-labels-and-target-snapshot" }, { "note": "Body states background execs, subagents, ACP sessions, and async events currently arrive as side-channel noise.", "type": "forgejo", "value": "issue-body-pain" }, { "note": "Acceptance direction keeps v0 read-only and MCP-backed rather than making Kan-ductor own process execution.", "type": "forgejo", "value": "issue-body-acceptance-direction" } ], "impact": 4, "judge_actor": { "name": "iskra", "runtime": "openclaw" }, "judged_at": "2026-06-10T01:08:00Z", "labels_to_apply": [ "judge/p1", "judge/operator-needed" ], "piotr_fit": "high", "priority": "p1", "rationale_summary": "This is P1 operator-shaped UX infrastructure because background work needs visible state and evidence to prevent stale replies and hidden promise debt.", "reach": 4, "recommended_next_action": "operator_needed", "rerun_reason": "no_prior_judgment", "schema": "openclaw.judge.v0", "target": { "kind": "issue", "number": 82, "repo": "pdurlej/kan-ductor" }, "target_snapshot": { "body_hash": "sha256:47cdd9f5d45977dba3b076bc13e41999b4384f26cc7498af61640d1663fe208a", "commit_count": null, "evidence_hash": "sha256:6f93e8c3e1d547404d307bb3d94e75d593dc4613978111835c151030a33970fa", "head_sha": null, "labels": [], "labels_hash": "sha256:e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "state": "open", "title_hash": "sha256:121173547054f7021712aa93912ce52620f9baf02dea75e237f37b984fac8b64", "updated_at": "2026-06-03T10:36:51+02:00" }, "top_caveat": "Keep it read-only and evidence-linked first; do not let Kan-ductor become a hidden process supervisor." }

Iskra added the

judge/operator-needed

judge/p1

labels

2026-06-10 03:09:20 +02:00