Dataset Viewer
Auto-converted to Parquet Duplicate
agent_persona
string
ambiguous_instructions
int64
artifacts_url
string
auditor_version
string
command_errors
int64
completed
bool
docs_url
string
failed_examples
int64
finding_count
int64
finished_at
string
grade
string
is_fixture
bool
missing_information
int64
model
string
navigation_operations
int64
provider
string
repository
string
rubric_version
string
run_id
string
scenario
string
score
int64
stale_instructions
int64
started_at
string
unsupported_assumptions
int64
valid
bool
validated
bool
new_developer
0
https://huggingface.co/datasets/critesjosh/agent-readiness-audits/tree/main/runs/ara_4abd7182116646708b731b0588b3e8b0
0.3.2
1
true
https://huggingface.co/docs/smolagents/index
0
0
2026-07-17T15:42:18.500023+00:00
A
false
0
Qwen/Qwen3-Coder-480B-A35B-Instruct
3
novita
https://github.com/huggingface/smolagents.git
2.0
ara_4abd7182116646708b731b0588b3e8b0
Hugging Face smolagents quickstart
98
0
2026-07-17T15:41:19.824525+00:00
0
true
true
new_developer
0
https://huggingface.co/datasets/critesjosh/agent-readiness-audits/tree/main/runs/ara_45e4a51aa7034f9eba88dae21bbe4c1e
0.3.3
0
false
https://docs.aztec.network/developers/testnet/getting_started_on_local_network
0
1
2026-07-17T16:21:29.738595+00:00
N/A
false
0
openai/gpt-oss-120b
8
cerebras
https://github.com/AztecProtocol/aztec-starter.git
2.0
ara_45e4a51aa7034f9eba88dae21bbe4c1e
Aztec local toolchain + contract scaffold (v5.0.0)
null
0
2026-07-17T16:19:02.580906+00:00
0
false
false
new_developer
0
https://huggingface.co/datasets/critesjosh/agent-readiness-audits/tree/main/runs/ara_8a96967c48d344a58f9c5ea925416601
0.3.3
0
true
https://docs.aztec.network/developers/ai_tooling
0
0
2026-07-17T16:56:42.130569+00:00
A
false
0
openai/gpt-oss-120b
2
cerebras
https://github.com/AztecProtocol/aztec-starter.git
2.0
ara_8a96967c48d344a58f9c5ea925416601
Aztec AI project guidance
100
0
2026-07-17T16:53:59.845655+00:00
0
true
true

Agent Readiness Audits

Public, reproducible results from the Agent Readiness Auditor. Each real run records the repository, target revision, scenario, model/provider, auditor and rubric versions, deterministic score, findings, and links to redacted Markdown/JSON artifacts.

The default split contains one compact, machine-readable row per live audit. The separate fixtures configuration contains a row explicitly marked is_fixture: true; it documents the schema and is not a measured claim about a real developer platform. Complete live artifacts are stored beneath runs/<run_id>/ and linked from each default-split row.

Privacy and trace policy

Only audits of public, allowlisted repositories may be published. Command output and reports pass through token-pattern redaction before upload. Raw chain-of-thought is never published. Tool actions, URLs, exit codes, errors, and evidence excerpts may be published after redaction.

Downloads last month
54

Collection including critesjosh/agent-readiness-audits