ultimate-pi 0.15.0 → 0.16.0

package/.agents/skills/harness-governor/SKILL.md

@@ -28,6 +28,17 @@ When refining plans from noisy requirements:
 3. When gates return `human_required` or promotion is blocked, the orchestrator calls `ask_user` — do not guess scope.
 4. Reference graphify wiki or `graphify query` for architecture constraints before execute.
 
+## Budgets (ADR 0038)
+
+- Default: **`HARNESS_BUDGET_ENFORCE` off** — token/debate caps are telemetry-only (`harness-budget-telemetry`, `harness-budget-soft-limit`). They do **not** block phases or debate lanes.
+- Do **not** skip scouts, debate rounds, or `approve_plan` because of soft budget hints in the widget.
+- Re-enable hard caps only with `HARNESS_BUDGET_ENFORCE=1` and `HARNESS_BUDGET_HARD_STOP` / `HARNESS_DEBATE_HARD_STOP`.
+
+## Subagent artifacts (ADR 0037)
+
+- Subagents call scoped **`submit_*`** tools; parent verifies with **`harness_artifact_ready`**, not JSON parsing from `finalOutput`.
+- Parent **`write_harness_yaml`** is for merges (`research-brief.yaml`, plan shell) — not subagent payloads.
+
 ## Rules
 
 - Never auto-merge; harness-auto may open PR only when all gates pass (see release-readiness-report).

package/.agents/skills/harness-orchestration/SKILL.md

@@ -14,6 +14,8 @@ description: >-
 
 Every spawn includes **HarnessSpawnContext** JSON in the task text (subprocess agents do not get `[HarnessActivePlan]` injection). Use `agentScope: "both"` so package agents under `$UP_PKG/.pi/agents/**` resolve.
 
+Harness subprocesses load **`harness-subagent-submit`** (`PI_HARNESS_SUBPROCESS=1`, `HARNESS_RUN_ID`, `HARNESS_RUN_DIR`). Agents must call their scoped **`submit_*`** tool before exit; parent gates use **`harness_artifact_ready`** and debate reads submit from `tool_result` (set `HARNESS_SUBMIT_TOOLS=0` only to fall back to `finalOutput` parsing).
+
 ## Subprocess telemetry
 
 Harness bridge emits `harness_subagent_spawned` / `harness_subagent_completed` (replaces in-process setup/blackboard events).
@@ -35,7 +37,7 @@ LIMIT 30
 
 1. **Parallel `tasks`** — one `subagent({ tasks: [...] })` for scouts, decompose+hypothesis, or review fan-in; subprocesses run in parallel upstream.
 2. **Blocking calls** — each `subagent` returns when the subprocess exits; no `get_subagent_result` polling.
-3. **Compact handoffs** — pass scout/decompose JSON only; never paste full subprocess message logs into the next spawn.
+3. **Compact handoffs** — read artifacts written by submit tools (or `harness_artifact_ready`); never paste full subprocess message logs into the next spawn.
 4. **No spawn cap** — harness subagent spawns are unlimited per session (active count is telemetry only). Do **not** pass `timeoutMs` unless the user wants a cap — subprocesses wait for natural exit (`PI_SUBAGENT_TIMEOUT_MS` optional env backstop only).
 
 ## Command → agent

package/.agents/skills/harness-plan/SKILL.md

@@ -11,14 +11,14 @@ description: PM-grade harness plans — scouts, Phase 3.5 implementation researc
 
 ## Workflow (parent orchestrator)
 
-1. Parallel scouts (graphify + structure; semantic unless `--quick`).
-2. Parallel decompose + hypothesis → `artifacts/decomposition.yaml`, `artifacts/hypothesis.yaml`.
-3. **Phase 3.5 (required):** parallel `implementation-researcher` + `stack-researcher` → `artifacts/implementation-research.yaml`, `artifacts/stack.yaml`; merge into `research-brief.yaml`.
+1. Parallel scouts (graphify + structure; semantic unless `--quick`) — each scout ends with **`submit_scout_findings`** (not JSON in final message).
+2. Parallel decompose + hypothesis — **`submit_decomposition`** / **`submit_hypothesis`**.
+3. **Phase 3.5 (required):** parallel `implementation-researcher` + `stack-researcher` — **`submit_implementation_research`** / **`submit_stack`**; parent merges into `research-brief.yaml` via `write_harness_yaml`.
 4. Draft `PlanPacket` shell; `ask_user` on material fork **after** Phase 3.5.
 5. `execution-plan-author` → merge `execution_plan`.
 6. **`validate-plan-dag.mjs`** (must pass).
-7. **`harness_plan_debate_eligibility`** → **`harness_debate_open`** with profile → Review Gate (required focuses per profile) → consensus.
-8. Apply patches, re-validate DAG, `approve_plan`, `create_plan`.
+7. **`harness_plan_debate_eligibility`** → **`harness_debate_open`** with profile → Review Gate (debate agents use lane **`submit_*`** tools; parent reads submit from `tool_result`, not `finalOutput` JSON).
+8. **`harness_artifact_ready`** on required paths → apply patches, re-validate DAG, `approve_plan`, `create_plan`.
 
 `--quick` skips semantic scout and post-run adversary only — **not** implementation research or plan debate.
 

package/.pi/agents/harness/adversary.md

@@ -1,6 +1,6 @@
 ---
 description: Adversarial harness reviewer focused on breaking assumptions and surfacing regressions.
-tools: read, grep, find, ls
+tools: read, grep, find, ls, submit_adversary_report
 extensions: false
 disallowed_tools: ask_user
 thinking: high

package/.pi/agents/harness/evaluator.md

@@ -1,6 +1,6 @@
 ---
 description: Independent harness evaluator producing structured pass/fail verdicts.
-tools: read, grep, find, ls
+tools: read, grep, find, ls, submit_eval_verdict
 extensions: false
 disallowed_tools: ask_user
 thinking: high

package/.pi/agents/harness/executor.md

@@ -1,6 +1,6 @@
 ---
 description: Harness executor that implements only within approved PlanPacket scope.
-tools: read, write, edit, bash, grep, find, ls
+tools: read, write, edit, bash, grep, find, ls, submit_executor_handoff
 extensions: true
 disallowed_tools: ask_user
 thinking: medium

package/.pi/agents/harness/incident-recorder.md

@@ -1,6 +1,6 @@
 ---
 description: Harness incident recorder compiling structured IncidentRecord drafts from run context.
-tools: read, grep, find, ls
+tools: read, grep, find, ls, submit_human_required
 extensions: false
 thinking: medium
 max_turns: 15

package/.pi/agents/harness/meta-optimizer.md

@@ -1,6 +1,6 @@
 ---
 description: Harness meta optimizer proposing policy/prompt/router improvements from trace evidence.
-tools: read, grep, find, ls
+tools: read, grep, find, ls, submit_human_required
 extensions: false
 disallowed_tools: ask_user
 thinking: high

package/.pi/agents/harness/planning/decompose.md

@@ -1,6 +1,6 @@
 ---
 description: Plan-phase DeepMind-style problem decomposition (read-only).
-tools: read, grep, find, ls, bash
+tools: read, grep, find, ls, bash, submit_decomposition_brief
 disallowed_tools: write, edit, ask_user, approve_plan, create_plan, subagent
 extensions: false
 thinking: medium
@@ -51,35 +51,6 @@ External / OSS prior art is **not** your job — `implementation-researcher` (Ph
 
 Identify contradictions, tradeoffs, or competing beliefs. Pick the **core tension** — one paragraph that feeds Phase 2 hypothesis generation.
 
-## Output (required JSON block)
-
-End with one fenced `json` block matching `PlanDecompositionBrief` (`.pi/harness/specs/plan-decomposition-brief.schema.json`):
-
-```json
-{
-  "schema_version": "1.0.0",
-  "problem_restatement": "…",
-  "problem_types": ["design"],
-  "scope": {
-    "narrowed_focus": "…",
-    "excluded": ["…"]
-  },
-  "hard_constraints": ["…"],
-  "soft_constraints": ["…"],
-  "success_metrics": ["…"],
-  "prior_art": {
-    "best_approach": "…",
-    "gap": "…",
-    "dead_ends": ["…"]
-  },
-  "tensions": [
-    {
-      "claim_a": "…",
-      "claim_b": "…",
-      "why_matters": "…"
-    }
-  ],
-  "core_tension": "…",
-  "human_summary": "…"
-}
-```
+## Output
+
+Before ending, call `submit_decomposition_brief` exactly once with the full `PlanDecompositionBrief` document. Do not paste the artifact as prose or a fenced JSON block — the tool write is the deliverable.

package/.pi/agents/harness/planning/execution-plan-author.md

@@ -1,6 +1,6 @@
 ---
 description: Plan-phase ExecutionPlan generator (PM-grade WBS + DAG).
-tools: read, grep, find, ls
+tools: read, grep, find, ls, submit_execution_plan_brief
 disallowed_tools: write, edit, bash, ask_user, approve_plan, create_plan, subagent
 extensions: false
 thinking: high
@@ -30,7 +30,8 @@ Task summary, `PlanDecompositionBrief`, `PlanHypothesisBrief`, draft scope/accep
 
 ## Output
 
-Valid **YAML only** — `PlanExecutionPlanBrief` with nested `execution_plan` (`.pi/harness/specs/plan-execution-plan-brief.schema.json`). Parent merges into `plan-packet.yaml` and runs `validate-plan-dag.mjs`.
+Before ending, call `submit_execution_plan_brief` exactly once with the full document. Prose summary is optional; the artifact is the tool call.
+
 
 ## Guardrails
 

package/.pi/agents/harness/planning/hypothesis-validator.md

@@ -1,6 +1,6 @@
 ---
 description: Plan-phase blind hypothesis validation (debate R1 only).
-tools: read, grep, find, ls
+tools: read, grep, find, ls, submit_hypothesis_validation
 disallowed_tools: write, edit, bash, ask_user, approve_plan, create_plan, subagent
 extensions: false
 thinking: medium
@@ -29,7 +29,8 @@ Ignore decomposition, scouts, PlanPacket, adversary output, prior debate rounds.
 
 ## Output
 
-Valid **YAML only** — `PlanHypothesisEval` (`.pi/harness/specs/plan-hypothesis-eval.schema.json`).
+Before ending, call `submit_hypothesis_validation` exactly once with the full document. Prose summary is optional; the artifact is the tool call.
+
 
 ## Guardrails
 

package/.pi/agents/harness/planning/hypothesis.md

@@ -1,6 +1,6 @@
 ---
 description: Plan-phase DARWIN hypothesis generation (read-only).
-tools: read, grep, find, ls, bash
+tools: read, grep, find, ls, bash, submit_hypothesis_brief
 disallowed_tools: write, edit, ask_user, approve_plan, create_plan, subagent
 extensions: false
 thinking: medium
@@ -61,29 +61,6 @@ Up to two alternatives with a different approach and **key_bet** (what it assume
 
 Do **not** include self-evaluation scores — a separate agent handles that.
 
-## Output (required JSON block)
-
-```json
-{
-  "schema_version": "1.0.0",
-  "primary": {
-    "claim": "…",
-    "mechanism": "…",
-    "prediction": "…",
-    "experiment": "…",
-    "tension_resolution": "…"
-  },
-  "dialectical_fork": {
-    "fork": "…",
-    "path_a": "…",
-    "path_b": "…"
-  },
-  "alternatives": [
-    { "claim": "…", "key_bet": "…" }
-  ],
-  "recommended_next_steps": ["…"],
-  "human_summary": "…"
-}
-```
-
-Match `PlanHypothesisBrief` (`.pi/harness/specs/plan-hypothesis-brief.schema.json`).
+## Output
+
+Before ending, call `submit_hypothesis_brief` exactly once with the full `PlanHypothesisBrief` document. Do not paste the artifact as prose or a fenced JSON block — the tool write is the deliverable.

package/.pi/agents/harness/planning/implementation-researcher.md

@@ -1,6 +1,6 @@
 ---
 description: Plan-phase external solution / prior-art research (web + in-repo, read-only writes via parent).
-tools: read, grep, find, ls, bash, web_search, web_fetch
+tools: read, grep, find, ls, bash, web_search, web_fetch, submit_implementation_research
 disallowed_tools: write, edit, ask_user, approve_plan, create_plan, subagent
 extensions: false
 thinking: medium
@@ -31,7 +31,8 @@ Read `HarnessSpawnContext` plus paths to `artifacts/decomposition.yaml`, `artifa
 
 ## Output
 
-Valid **YAML only** (no markdown fences) — `PlanImplementationResearchBrief` (`.pi/harness/specs/plan-implementation-research-brief.schema.json`). Parent writes `artifacts/implementation-research.yaml`.
+Before ending, call `submit_implementation_research` exactly once with the full document. Prose summary is optional; the artifact is the tool call.
+
 
 ## Guardrails
 

package/.pi/agents/harness/planning/plan-adversary.md

@@ -1,6 +1,6 @@
 ---
 description: Plan-phase adversarial verification on ExecutionPlan.
-tools: read, grep, find, ls
+tools: read, grep, find, ls, submit_adversary_brief
 disallowed_tools: write, edit, bash, ask_user, approve_plan, create_plan, subagent
 extensions: false
 thinking: medium
@@ -21,9 +21,8 @@ Stress-test the ExecutionPlan with reproducible counterexamples. Map every findi
 
 ## Output
 
-Valid **YAML only** — `PlanAdversaryBrief` (`.pi/harness/specs/plan-adversary-brief.schema.json`).
+Before ending, call `submit_adversary_brief` exactly once with the full document. Prose summary is optional; the artifact is the tool call.
 
-Include `open_claim_ids: string[]` for claims still disputed after your message (parent tracks ping-pong).
 
 ## Guardrails
 

package/.pi/agents/harness/planning/plan-evaluator.md

@@ -1,6 +1,6 @@
 ---
 description: Plan-phase Validation Checks evaluator (neutral pass/fail).
-tools: read, grep, find, ls
+tools: read, grep, find, ls, submit_validation_turn
 disallowed_tools: write, edit, bash, ask_user, approve_plan, create_plan, subagent
 extensions: false
 thinking: medium
@@ -30,7 +30,8 @@ Parent passes `debate_round_focus`: `spec` | `wbs` | `schedule` | `quality`. Use
 
 ## Output
 
-Valid **YAML only** — `PlanValidationTurn` (`.pi/harness/specs/plan-validation-turn.schema.json`). Fail the round in output if `dag_validation.status === "fail"` when visible in packet.
+Before ending, call `submit_validation_turn` exactly once with the full document. Prose summary is optional; the artifact is the tool call.
+
 
 ## Guardrails
 

package/.pi/agents/harness/planning/review-integrator.md

@@ -1,6 +1,6 @@
 ---
 description: Plan-phase Review Gate integrator (round → debate bus).
-tools: read, grep, find, ls
+tools: read, grep, find, ls, submit_review_round_draft
 disallowed_tools: write, edit, bash, ask_user, approve_plan, create_plan, subagent
 extensions: false
 thinking: medium
@@ -26,9 +26,8 @@ Synthesize evaluator, adversary, sprint audit, and (R1) hypothesis-validator lan
 
 ## Output
 
-Valid **YAML only** — `PlanReviewRoundDraft` (`.pi/harness/specs/plan-review-round-draft.schema.json`) including `debate_round_focus`.
+Before ending, call `submit_review_round_draft` exactly once with the full document. Prose summary is optional; the artifact is the tool call.
 
-Parent calls `harness_debate_submit_round` — you do not write `review-round-r*.yaml` yourself.
 
 ## Guardrails
 

package/.pi/agents/harness/planning/scout-graphify.md

@@ -1,6 +1,6 @@
 ---
 description: Plan-phase scout — graphify graph and wiki navigation (read-only).
-tools: read, bash, ls
+tools: read, bash, ls, submit_scout_findings
 disallowed_tools: write, edit, ask_user, approve_plan, create_plan, subagent, grep, find
 extensions: false
 thinking: low
@@ -32,25 +32,6 @@ Read `HarnessSpawnContext` in the spawn prompt (`task_summary`, `mode`, `plan_pa
 
 Read-only only: no `graphify update`, `graphify extract`, `pip install`, redirects (`>`, `>>`), or file creation. Allowed: `graphify query`, `graphify path`, `graphify explain`, `ls`, `cat`, `head`.
 
-## Output limits
+## Output
 
-- `findings`: at most **8** bullets, each ≤2 sentences
-- `key_paths`: at most **10** absolute paths
-- `open_questions`: at most **5** items
-
-## Output (required JSON block)
-
-End with one fenced `json` block:
-
-```json
-{
-  "schema_version": "1.0.0",
-  "lane": "graphify",
-  "status": "ok",
-  "findings": ["…"],
-  "key_paths": ["/absolute/path"],
-  "open_questions": ["…"]
-}
-```
-
-Use `"status": "partial"` if the graph is missing or queries failed; still return best-effort findings.
+Before ending, call `submit_scout_findings` exactly once with the full document (`schema_version`, `lane`, `status`, `findings`, `key_paths`, `open_questions`). Use `"status": "partial"` if the graph is missing or queries failed. Do not paste the artifact as prose — the tool write is the deliverable.

package/.pi/agents/harness/planning/scout-semantic.md

@@ -1,6 +1,6 @@
 ---
 description: Plan-phase scout — CocoIndex semantic code search (read-only).
-tools: read, bash, ls
+tools: read, bash, ls, submit_scout_findings
 disallowed_tools: write, edit, ask_user, approve_plan, create_plan, subagent, grep, find
 extensions: false
 thinking: low
@@ -34,21 +34,6 @@ Read-only only: no installs, indexing, daemon control, or redirects.
 
 **Forbidden:** `ccc index`, `ccc init`, `ccc reset`, `ccc daemon`, `ccc search --refresh`, package installs.
 
-## Output limits
+## Output
 
-- `findings`: at most **6** bullets
-- `key_paths`: at most **8** absolute paths
-- `open_questions`: at most **4** items
-
-## Output (required JSON block)
-
-```json
-{
-  "schema_version": "1.0.0",
-  "lane": "semantic",
-  "status": "ok",
-  "findings": ["…"],
-  "key_paths": ["/absolute/path"],
-  "open_questions": ["…"]
-}
-```
+Before ending, call `submit_scout_findings` exactly once with the full document (`schema_version`, `lane`, `status`, `findings`, `key_paths`, `open_questions`). Do not paste the artifact as prose — the tool write is the deliverable.

package/.pi/agents/harness/planning/scout-structure.md

@@ -1,6 +1,6 @@
 ---
 description: Plan-phase scout — ast-grep structural code search (read-only).
-tools: read, bash, ls
+tools: read, bash, ls, submit_scout_findings
 disallowed_tools: write, edit, ask_user, approve_plan, create_plan, subagent, grep, find
 extensions: false
 thinking: low
@@ -30,21 +30,6 @@ Read `HarnessSpawnContext` in the spawn prompt. For `mode: revise`, read the exi
 
 Read-only only: no installs, redirects, or mutating git/npm commands.
 
-## Output limits
+## Output
 
-- `findings`: at most **8** bullets
-- `key_paths`: at most **10** absolute paths
-- `open_questions`: at most **5** items
-
-## Output (required JSON block)
-
-```json
-{
-  "schema_version": "1.0.0",
-  "lane": "structure",
-  "status": "ok",
-  "findings": ["…"],
-  "key_paths": ["/absolute/path"],
-  "open_questions": ["…"]
-}
-```
+Before ending, call `submit_scout_findings` exactly once with the full document (`schema_version`, `lane`, `status`, `findings`, `key_paths`, `open_questions`). Do not paste the artifact as prose — the tool write is the deliverable.

package/.pi/agents/harness/planning/sprint-contract-auditor.md

@@ -1,6 +1,6 @@
 ---
 description: Plan-phase ADR-020 sprint contract auditor.
-tools: read, grep, find, ls
+tools: read, grep, find, ls, submit_sprint_audit
 disallowed_tools: write, edit, bash, ask_user, approve_plan, create_plan, subagent
 extensions: false
 thinking: medium
@@ -23,7 +23,8 @@ Required when `debate_round_focus` is `quality` or round_index ≥ 4. Optional s
 
 ## Output
 
-Valid **YAML only** — `PlanSprintAuditTurn` (`.pi/harness/specs/plan-sprint-audit-turn.schema.json`).
+Before ending, call `submit_sprint_audit` exactly once with the full document. Prose summary is optional; the artifact is the tool call.
+
 
 ## Guardrails
 

package/.pi/agents/harness/planning/stack-researcher.md

@@ -1,6 +1,6 @@
 ---
 description: Plan-phase stack research (ctx7 + web, read-only file writes via parent).
-tools: read, grep, find, ls, bash, web_search, web_fetch
+tools: read, grep, find, ls, bash, web_search, web_fetch, submit_stack_brief
 disallowed_tools: write, edit, ask_user, approve_plan, create_plan, subagent
 extensions: false
 thinking: medium
@@ -22,7 +22,8 @@ Produce evidence-backed stack recommendations before ExecutionPlan authoring. Ra
 
 ## Output
 
-Valid **YAML only** (no markdown fences) — `PlanStackBrief` (`.pi/harness/specs/plan-stack-brief.schema.json`). Parent writes `artifacts/stack.yaml`.
+Before ending, call `submit_stack_brief` exactly once with the full document. Prose summary is optional; the artifact is the tool call.
+
 
 ## Guardrails
 

package/.pi/agents/harness/tie-breaker.md

@@ -1,6 +1,6 @@
 ---
 description: Final arbiter for unresolved evaluator vs adversary debates within budget limits.
-tools: read, grep, find, ls
+tools: read, grep, find, ls, submit_human_required
 extensions: false
 disallowed_tools: ask_user
 thinking: high

package/.pi/agents/harness/trace-librarian.md

@@ -1,6 +1,6 @@
 ---
 description: Harness trace librarian for run replay, artifact indexing, and forensics summaries.
-tools: read, grep, find, ls
+tools: read, grep, find, ls, submit_human_required
 extensions: false
 thinking: medium
 max_turns: 20

package/.pi/extensions/budget-guard.ts

@@ -8,6 +8,10 @@
 import { appendFile, mkdir, readFile } from "node:fs/promises";
 import { join } from "node:path";
 import type { ExtensionAPI } from "@earendil-works/pi-coding-agent";
+import {
+	isHarnessBudgetEnforceOn,
+	shouldEmitBlockingBudgetExhausted,
+} from "../lib/harness-budget-enforce.js";
 import { getRunIdFromSession } from "../lib/harness-run-context.js";
 
 type HarnessPhase = "plan" | "execute" | "evaluate" | "adversary" | "merge";
@@ -52,7 +56,8 @@ const EVENTS_FILE = join(RUNS_DIR, "budget-events.jsonl");
 const DEFAULT_GLOBAL_CAP = Number(
 	process.env.HARNESS_BUDGET_TOTAL_TOKENS ?? "120000",
 );
-const HARD_STOP_BUDGETS = process.env.HARNESS_BUDGET_HARD_STOP === "true";
+const HARD_STOP_BUDGETS =
+	process.env.HARNESS_BUDGET_HARD_STOP === "true" && isHarnessBudgetEnforceOn();
 const DEFAULT_PHASE_CAPS: Record<HarnessPhase, number> = {
 	plan: Number(process.env.HARNESS_BUDGET_PLAN_TOKENS ?? "80000"),
 	execute: Number(process.env.HARNESS_BUDGET_EXECUTE_TOKENS ?? "80000"),
@@ -190,7 +195,9 @@ async function emitBudgetEvent(
 	await ensureRunsDir();
 	const line = `${JSON.stringify({ timestamp: nowIso(), ...event })}\n`;
 	await appendFile(EVENTS_FILE, line, "utf-8");
-	pi.appendEntry("harness-budget-exhausted", event);
+	if (shouldEmitBlockingBudgetExhausted()) {
+		pi.appendEntry("harness-budget-exhausted", event);
+	}
 }
 
 const debouncedSoftLimit = new Map<string, boolean>();
@@ -240,26 +247,33 @@ export default function budgetGuard(pi: ExtensionAPI) {
 		};
 
 		const debounceKey = `${runId}:${phase}:${exhaustionReason}`;
-		if (!debouncedSoftLimit.has(debounceKey)) {
-			debouncedSoftLimit.set(debounceKey, true);
-			await emitBudgetEvent(pi, exhausted);
+		const softKey = `${debounceKey}:soft`;
+		if (!debouncedSoftLimit.has(softKey)) {
+			debouncedSoftLimit.set(softKey, true);
+			pi.appendEntry("harness-budget-soft-limit", {
+				run_id: exhausted.run_id,
+				phase,
+				phaseUsed,
+				phaseCap,
+				totalUsed: usage.totalTokens,
+				totalCap: globalCap,
+				exhaustion_reason: exhaustionReason,
+				timestamp: nowIso(),
+			});
+			pi.appendEntry("harness-budget-telemetry", {
+				...exhausted,
+				telemetry_only: !isHarnessBudgetEnforceOn(),
+			});
 		}
 
-		if (!HARD_STOP_BUDGETS) {
-			const softKey = `${debounceKey}:soft`;
-			if (!debouncedSoftLimit.has(softKey)) {
-				debouncedSoftLimit.set(softKey, true);
-				pi.appendEntry("harness-budget-soft-limit", {
-					run_id: exhausted.run_id,
-					phase,
-					phaseUsed,
-					phaseCap,
-					totalUsed: usage.totalTokens,
-					totalCap: globalCap,
-					exhaustion_reason: exhaustionReason,
-					timestamp: nowIso(),
-				});
+		if (isHarnessBudgetEnforceOn()) {
+			if (!debouncedSoftLimit.has(debounceKey)) {
+				debouncedSoftLimit.set(debounceKey, true);
+				await emitBudgetEvent(pi, exhausted);
 			}
+		}
+
+		if (!HARD_STOP_BUDGETS) {
 			return undefined;
 		}
 		return {

package/.pi/extensions/harness-debate-tools.ts

@@ -8,6 +8,10 @@ import type { ExtensionAPI } from "@earendil-works/pi-coding-agent";
 import { Type } from "@sinclair/typebox";
 import { parse as parseYaml } from "yaml";
 import type { DebateParticipant } from "../lib/debate-orchestrator-types.js";
+import {
+	extractLastSubmitCall,
+	type MessageLike,
+} from "../lib/harness-agent-output.js";
 import {
 	getLatestRunContext,
 	getRunIdFromSession,
@@ -22,6 +26,7 @@ import {
 import { getDebateState } from "./lib/debate-bus-state.js";
 import { claimExtensionLoad } from "./lib/extension-load-guard.js";
 import { captureHarnessEvent } from "./lib/harness-posthog.js";
+import { DEBATE_AGENT_SUBMIT_TOOL } from "./lib/harness-subagent-submit-registry.js";
 import {
 	type DebateEligibilityInput,
 	harnessPlanDebateEligibility,
@@ -40,6 +45,7 @@ import {
 } from "./lib/plan-debate-id.js";
 import {
 	applyDebateLane,
+	applyDebateLaneFromDoc,
 	type DebateLaneKind,
 	debateLaneForAgent,
 	formatApplyLaneMessage,
@@ -95,13 +101,19 @@ function telemetryRound(
 
 function subagentResults(
 	details: unknown,
-): Array<{ agent: string; finalOutput?: string }> {
+): Array<{ agent: string; finalOutput?: string; messages?: MessageLike[] }> {
 	const d = details as {
-		results?: Array<{ agent: string; finalOutput?: string }>;
+		results?: Array<{
+			agent: string;
+			finalOutput?: string;
+			messages?: MessageLike[];
+		}>;
 	};
 	return d?.results ?? [];
 }
 
+const USE_SUBMIT_TOOLS = process.env.HARNESS_SUBMIT_TOOLS !== "0";
+
 export default function harnessDebateTools(pi: ExtensionAPI) {
 	if (!claimExtensionLoad("harness-debate-tools", MODULE_URL)) return;
 
@@ -118,7 +130,34 @@ export default function harnessDebateTools(pi: ExtensionAPI) {
 		let lastRound = 1;
 		for (const result of subagentResults(event.details)) {
 			const lane = debateLaneForAgent(result.agent ?? "");
-			if (!lane || !result.finalOutput?.trim()) continue;
+			if (!lane) continue;
+
+			const submitTool = DEBATE_AGENT_SUBMIT_TOOL[result.agent ?? ""];
+			const submitCall =
+				USE_SUBMIT_TOOLS && submitTool && result.messages
+					? extractLastSubmitCall(result.messages, submitTool)
+					: null;
+
+			if (submitCall) {
+				const out = await applyDebateLaneFromDoc({
+					runDir: rd,
+					lane,
+					doc: submitCall.document,
+				});
+				if (out.round_index) lastRound = out.round_index;
+				pi.appendEntry("harness-debate-lane-applied", {
+					agent: result.agent,
+					source: "submit_tool",
+					tool: submitCall.toolName,
+					...out,
+				});
+				applied.push(formatApplyLaneMessage(out));
+				continue;
+			}
+
+			if (!result.finalOutput?.trim()) continue;
+			if (USE_SUBMIT_TOOLS && submitTool) continue;
+
 			const out = await applyDebateLane({
 				runDir: rd,
 				lane,