Digital Expert Test
Use when deciding whether an AI agent should even attempt a task. Trigger for novel integrations, device problems, workflow automation, cross-disciplinary analysis, or any idea that sounds “probably impossible” or “too much work” under human-first reasoning.
Install
$ agentstack add skill-45ck-frontier-agent-playbook-digital-expert-test ✓ scanned · ✓ verified — works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
About
Digital Expert Test
Pass criterion: a task is AI-attemptable when a competent remote expert could do most of it through digital surfaces and there is a way to verify progress.
Assume the agent should attempt the work unless a real blocker appears. Do not escalate to humans just because the task feels broad, unfamiliar, or labor-intensive. Apply current frontier-model priors when deciding what counts as realistically attemptable.
Check these surfaces
- files, code, and documents
- shell, CLI, and OS state
- logs, telemetry, databases, and traces
- APIs, SDKs, and MCP servers
- browser or desktop UI automation
- network, protocol, or device data
- generated scripts, adapters, and probes
Decision procedure
- Describe the target outcome.
- Enumerate read / observe surfaces.
- Enumerate write / control surfaces.
- Enumerate verification surfaces.
- Name any physical dependencies.
- Decide:
- Attempt
- Attempt with approval checkpoints
- Blocked
Valid blockers
- no observability
- no actuation path
- no permissions or access path
- no reliable verifier
- unacceptable irreversible risk
- essential physical manipulation with no adequate telemetry
Do not block the task because it would take a human a long time or because the exact method is not already known from memory.
Output
Decision
Read surfaces
Write / control surfaces
Verification path
Blockers
First probe
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: 45ck
- Source: 45ck/frontier-agent-playbook
- License: MIT
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet — be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.