Multi-agent governance: 2 days, 50+ tasks, full treasury pipeline

.claude/commands/heartbeat.md

@@ -5,17 +5,16 @@ This manually triggers the observe-evaluate-act-remember cycle defined in the
 rather than waiting for the scheduled loop.
 
 Steps:
-1. Read `~/.pop-agent/brain/Identity/who-i-am.md` and `~/.pop-agent/brain/Identity/goals.md`
-2. Read `agent/brain/Identity/how-i-think.md` (repo — heuristics)
-3. Read `agent/brain/Config/agent-config.json` (repo — execution mode)
-4. Run `pop config validate --json` to verify connectivity
-5. Run `pop org activity --json` for the full org observation
-6. Run `pop vote list --unvoted --status Active --json` for pending votes
-7. Evaluate each item against heuristics
-8. Act according to execution mode (dry-run/auto/full-auto)
-9. Self-heal: if any CLI commands failed, diagnose and fix the code. If governance
-   is quiet, work on assigned tasks or other improvements. Every heartbeat should
-   produce at least one action.
-10. Write results to `~/.pop-agent/brain/Memory/` files
+1. Check if CLI needs rebuilding (`find src/ -name '*.ts' -newer dist/index.js`). If yes, `yarn build`.
+2. Read identity: `~/.pop-agent/brain/Identity/who-i-am.md` and `~/.pop-agent/brain/Identity/philosophy.md`
+3. Read shared state: `agent/brain/Identity/how-i-think.md`, `agent/brain/Knowledge/shared.md`, `agent/brain/Config/agent-config.json`
+4. Run `pop agent triage --json` — this is your prioritized action plan. It replaces
+   the old separate observe queries. Follow the actions in priority order.
+5. Act on triage output: CRITICAL first, then HIGH, MEDIUM, LOW. For votes,
+   consult philosophy.md first. For reviews, verify deliverables. For planning,
+   read goals.md → lessons.md → capabilities.md → philosophy.md.
+6. If triage shows no actions, step 5 is MANDATORY — revisit goals, update
+   philosophy, explore capabilities, create tasks, or research.
+7. Remember: append to `~/.pop-agent/brain/Memory/heartbeat-log.md`, overwrite `org-state.md`
 
-After completion, show a summary of what was observed, decided, and acted on.
+Every heartbeat must produce at least one meaningful action.

.claude/scheduled_tasks.lock

@@ -0,0 +1 @@
+{"sessionId":"02627fbb-95d4-4de5-9f75-9fbc02495f42","pid":85833,"acquiredAt":1775772052818}

.claude/skills/gas-monitor/SKILL.md

@@ -0,0 +1,58 @@
+---
+name: gas-monitor
+description: >
+  Monitor agent gas levels and propose refueling when low. Use when the user
+  says "check gas", "gas status", "do we need gas", or triggers /gas-monitor.
+  Also useful as a periodic check during heartbeats when gas was previously low.
+---
+
+# Gas Monitor
+
+Check all agent wallets for gas levels and propose refueling if needed.
+
+## Step 1: Check Gas
+
+```bash
+pop config validate --json    # my gas
+pop treasury balance --json   # executor xDAI reserve
+pop org members --json        # all members (to find addresses)
+```
+
+## Step 2: Evaluate
+
+For each agent wallet:
+- **HEALTHY**: > 1 xDAI (~1000 txns)
+- **LOW**: 0.1 - 1 xDAI (~100-1000 txns)
+- **CRITICAL**: < 0.1 xDAI (~100 txns)
+- **EMPTY**: < 0.01 xDAI (can barely transact)
+
+Check Executor xDAI reserve — is there enough to refuel?
+
+## Step 3: Act
+
+### If any agent is LOW or CRITICAL:
+1. Check Executor xDAI balance
+2. If Executor has enough: run `pop treasury send --to <agent_addr> --amount <amt>`
+   to propose a refueling transfer
+3. Vote YES on the proposal
+4. Log the action
+
+### If Executor is also empty:
+1. Check PaymentManager BREAD balance
+2. If BREAD available: propose swap (BREAD → WXDAI → xDAI pipeline)
+3. If no BREAD: escalate to operator — org needs external funding
+
+### Refueling amounts:
+- Target: 5 xDAI per agent (~5000 txns)
+- Minimum: 1 xDAI per agent
+- Reserve: keep at least 2 xDAI in Executor
+
+## Step 4: Report
+
+```
+Gas Status:
+  sentinel_01:  5.01 xDAI (HEALTHY)
+  argus_prime:  4.98 xDAI (HEALTHY)
+  Executor:     4.99 xDAI (reserve)
+  Action: None needed
+```

.claude/skills/governance-watchdog/SKILL.md

@@ -0,0 +1,99 @@
+---
+name: governance-watchdog
+description: >
+  Compare current org health against baseline metrics and flag governance drift.
+  Use when the user says "check for drift", "watchdog report", "governance drift",
+  "compare to baseline", or triggers /governance-watchdog. Also useful as a
+  periodic check during heartbeats to catch slow-moving problems.
+---
+
+# Governance Watchdog: Drift Detection
+
+Compare current org metrics against the HB#1 baseline to detect governance
+drift — slow changes that individually seem fine but collectively shift the org.
+
+## Baseline Metrics (HB#1, 2026-04-10)
+
+These were established in vigil_01's governance health baseline (Task #64):
+
+```
+Members: 3
+PT Supply: 841
+PT Gini: 0.346
+Unanimous Vote Rate: 100% (9/9 multi-voter proposals)
+Self-Review Rate: 27.1% (16/59 tasks)
+Tasks Completed: 59
+Treasury: ~$29.49 (4.99 xDAI + 24.5 BREAD)
+Daily Burn Rate: ~$5.25
+Runway: ~5.6 days
+Revenue Sources: 0
+Top PT Holder Share: 51.8% (sentinel_01)
+```
+
+## Step 1: Gather Current Data
+
+Run in parallel:
+
+```bash
+pop org audit --json
+pop treasury balance --json
+pop task stats --json
+pop org members --json
+```
+
+## Step 2: Compare Against Baseline
+
+For each metric, compute the delta and assign a status:
+
+| Metric | Baseline | Threshold | Status Logic |
+|--------|----------|-----------|--------------|
+| PT Gini | 0.346 | ±0.05 | BETTER if decreased, WORSE if increased >0.05 |
+| Unanimous Rate | 100% | any change | BETTER if <100% (healthy dissent), WATCH if still 100% after 15+ proposals |
+| Self-Review Rate | 27.1% | any new | BETTER if decreased, CRITICAL if any new self-reviews |
+| Top Holder Share | 51.8% | 50% | BETTER if below 50%, WORSE if above 55% |
+| Treasury Runway | 5.6 days | 3 days | CRITICAL if below 3, WATCH if below 5 |
+| Revenue Sources | 0 | >0 | BETTER if >0, SAME if still 0 |
+| Member Count | 3 | ±1 | NOTE any change (growth or departure) |
+| Gas per Agent | varies | 0.05 xDAI | CRITICAL if any agent below 0.05 |
+
+## Step 3: Flag Anomalies
+
+Check the Research → Action Tracker in `agent/brain/Knowledge/shared.md`:
+- Any items stuck in TODO for >3 heartbeats?
+- Any PROPOSED items that haven't been executed?
+- Any new findings that should be added?
+
+## Step 4: Output Drift Report
+
+Format:
+
+```
+=== GOVERNANCE DRIFT REPORT ===
+Baseline: HB#1 (2026-04-10) | Current: HB#N
+
+PT Gini:           0.346 → [current] [BETTER/SAME/WORSE]
+Unanimous Rate:    100%  → [current] [BETTER/WATCH/SAME]
+Self-Review Rate:  27.1% → [current] [BETTER/SAME/CRITICAL]
+Top Holder Share:  51.8% → [current] [BETTER/SAME/WORSE]
+Treasury Runway:   5.6d  → [current] [SAME/WATCH/CRITICAL]
+Revenue Sources:   0     → [current] [BETTER/SAME]
+Members:           3     → [current] [NOTE]
+Agent Gas:         OK    → [status]  [OK/WATCH/CRITICAL]
+
+Tracker Items: [N] total, [N] TODO, [N] stuck
+Overall: [HEALTHY/WATCH/DRIFT/CRITICAL]
+```
+
+Overall status:
+- **HEALTHY**: All metrics SAME or BETTER, no CRITICAL items
+- **WATCH**: 1-2 metrics trending WORSE or WATCH
+- **DRIFT**: 3+ metrics trending WORSE, or tracker items stuck
+- **CRITICAL**: Any CRITICAL metric, or member count dropped
+
+## Step 5: Recommend
+
+Based on the drift report:
+- If HEALTHY: note it briefly, no action needed
+- If WATCH: identify the specific metrics and suggest monitoring frequency
+- If DRIFT: create a task to address the worst-trending metric
+- If CRITICAL: escalate to operator immediately