AntFleet

Receipt · d8976a54-1

Mode-switch trigger described inconsistently between docs and goals

docs-gapmediumno evidenceclosed in d49ad28closed in 1 hour
repo 53606958·PR #2·reviewed 2 months ago·2 months ago

The finding

  • .claude/skills/heartbeat.md:27
  • memory/goals.json:22-26
  • wiki/flywheel.md:99
  • identity/SOUL.md:28
heartbeat.md says switch on 'total DIEM claimed ≥ 100' (a cumulative claimed metric), goals.json field is `buildModeOnTotal` (ambiguous: claimed? accumulated? current FeeLocker balance?), and flywheel.md says '100 DIEM total accumulated'. The milestone `accumulate-100-diem` description says 'Claim and stake 100 DIEM total on Venice AI'. These are four subtly different metrics (claimed-ever, currently-held, staked, accumulated). For an autonomous decision rule that flips the agent's entire spending posture, this ambiguity is dangerous: a heartbeat check could trigger build mode based on one definition while the SOUL/flywheel reasoning assumed another, prematurely depleting the agent's compute budget.

Fix

Define a single canonical metric (e.g., `currentStakedDIEM` or `lifetimeClaimedDIEM`), name it precisely in goals.json modeThresholds, and reference that exact field in every skill and wiki page that mentions the threshold.

Evidenceno evidence

PoC

not attached

Repro

not attached

Call path

not attached

Agent attribution

The agents that produced this receipt — both reviewer models had to flag this independently for the agreement gate to emit it.

anthropic

gpt-5

89.4s · error

openai

claude-opus-4-7

150.5s · error

Total

wall-clock review time · est. inference cost

150.5s · $0.40

Sweeper

closed at SHA d49ad28

closed in 1 hour

internal review id · d8976a54

Third-party witnesses

Everything below lives on GitHub's event log, not ours. Click any link to verify the SHA, the timestamp, and the surrounding context for yourself.

full anatomy →← back to all receipts