Trust ledger
Operations riskAggregate risk and reliability across all runs.
What this is
Aggregated reliability across every mission, crew, tool, memory, and approval. The ledger reads from receipts — if no receipts exist yet, this page is empty.
What to do
Scan the KPI tiles for red. Click any high-risk row to open its receipt and see what triggered the score.
What happens next
Use the trends to tune memory, tighten prompts, or block a flaky tool. Receipts back every number with a verifiable hash.
Missions
7
Receipts
4 (1 final)
Approval rate
50%
Failed jobs
0
Tool failures
3
Blocked tools
0
Memory accepted
2
Memory rejected
0
Hygiene flags
2
Hermes memory
online
High risk
0
Medium risk
2
Receipt coverage
57% of missions have workflow receipts. Finalized receipts are locked proof; drafts can still be regenerated.
Recent high-risk receipts
Tap any row to open its receipt
Most used tools
Across all missions
terminal_exec
7
price_series_lookup
4
deadline_create
2
web_snapshot
1
Hermes runtime · sandbox
terminal_exec + python_rpc across all missions
Total
7
Completed
7
Blocked
0
Failed
0
terminal_exec7 calls
docker 0local 7
python_rpc0 calls
docker 0local 0
Approvals by type
REPORT_EXPORT2 ok0 rej1 pendtotal 3
TRADE_SIMULATION2 ok0 rej1 pendtotal 3
EMAIL_DRAFT0 ok0 rej1 pendtotal 1
FOLLOW_UP0 ok0 rej1 pendtotal 1
Most injected memories
5×GLOBALHERMES[demo] Operator timezone: Europe/London (UK)
3×CREWLOCALApproved paper-trade structure
2×CREWLOCAL[demo] Always state simulation only on paper trades
2×CREWLOCAL[demo] Bug Hunter — concise bullets in reports
2×CREWLOCAL[demo] Use verbose, detailed bug reports
1×CREWLOCAL[demo] Prefer calm, firm tone for refund disputes
Crews by reliability
Bug Hunter Crew1 runsfail 0%
Life Admin Crew1 runsfail 0%
Paper Trading Desk2 runsfail 0%
Memory changes (last 14 days)
18
19
20
21
22
23
24
25
26
27
28
29
30
1
Browser QA · Bug Hunter Crew
Real Playwright audits + synthetic fallbacks
Pages audited
0
High issues
0
Medium issues
0
Low issues
0
Fallbacks
0
Failures
0
No browser audits yet. Run a Bug Hunter mission against /demo-target (set
ALLOW_LOCAL_BROWSER_QA=true in dev) or any public URL.