harn-cli 0.8.36

CLI for the Harn programming language — run, test, REPL, format, and lint
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34

# Seed evidence for local coding-agent model recommendations.
# `harn eval coding-agent --include-local` writes a fresher
# `.harn-runs/coding-agent-bench/latest/local_readiness.json`; this file is
# only the fallback when no local benchmark report exists yet.

[[outcomes]]
provider = "ollama"
model = "devstral-small-2:24b"
selector = "devstral-small-2"
tool_format = "text"
class = "passed"
status = "passed"
score = 100
evidence = "First empirical coding-agent fixture passed in text-tool mode."

[[outcomes]]
provider = "ollama"
model = "qwen3.6:35b-a3b-coding-nvfp4"
selector = "qwen3.6-coding"
tool_format = "text"
class = "provider_transport_failure"
status = "infra_error"
score = 35
evidence = "Initial local run hit an Ollama HTTP 500 / EOF transport path; report separately from model task quality."

[[outcomes]]
provider = "ollama"
model = "gemma4:26b"
selector = "ollama-gemma4"
tool_format = "text"
class = "behavioral_failure"
status = "failed"
score = 20
evidence = "Initial local run failed behaviorally by hallucinating paths/tools in the coding-agent fixture."