All sessions

kimi k3, bare agent

completed

key_limit 12 because output runs $15/M, the dearest in this sweep: AFI_MAX_TOKENS 64000 reserves $0.96, and the key has to hold that on top of what a run has already spent / Started 31 Aug 2026, 16:13

Started
Completed
Planned3
Produced3
Missing0

Plan completion

Every requested trial remains visible, including runs that produced no record

SequenceReviewerRun identityLifecycleReasonInspect run
1afiafi/kimi-k3/20260831T161340Z completed-View evidence
2afiafi/kimi-k3/20260831T161341Z completed-View evidence
3afiafi/kimi-k3/20260831T161342Z completed-View evidence

Produced runs

View evidence opens one run detail

Details
afi
afi · cli
default Default profile · Moonshot AI · kimi k3 high
proxy30%N/A7 scoredView evidence
afi
afi · cli
default Default profile · Moonshot AI · kimi k3 high
proxy35%N/A7 scoredView evidence
afi
afi · cli
default Default profile · Moonshot AI · kimi k3 high
proxy40%N/A8 scoredView evidence

Session specification

The committed JSON is shown as data, not interpreted as instructions

{
  "defaults": {
    "effort": "high",
    "key_limit": 12,
    "mode": "agent",
    "pr": "smykla-skalski/cairn#8",
    "subject": "proxy",
    "timeout": 900
  },
  "description": "key_limit 12 because output runs $15/M, the dearest in this sweep: AFI_MAX_TOKENS 64000 reserves $0.96, and the key has to hold that on top of what a run has already spent.",
  "name": "kimi k3, bare agent",
  "runs": [
    {
      "label": "kimi-k3",
      "model": "openrouter:moonshotai/kimi-k3",
      "repeat": 3
    }
  ]
}

Session evidence

Committed specification and session record

record present

Open verified source artifact

Open this path at the exported commit

The artifact is present; its content is available through the verified source link

spec present

Open verified source artifact

Open this path at the exported commit

The artifact is present; its content is available through the verified source link

benchee benchee-dashboard-1 built from 10f4ec58 Static benchmark evidence ·