Content
82%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A dense, well-structured adversary-emulation playbook that maps each kill-chain phase to concrete actions and delegated skills, with strong OPSEC, RoE, and deconfliction guidance and no wasted tokens. Main gaps are that execution relies on pointers to external skills rather than inline executable code, and validation is framed as safety gates rather than technical feedback loops.
Suggestions
Add a short validate→fix→retry feedback loop for the technical artifacts the skill does produce (e.g., validate the JSON threat-profile/conops against its schema before recording), which would lift workflow_clarity to a 5.
For the highest-risk step (edge-device exploitation), make the pre-exploit validation an explicit ordered checklist (confirm authorization → confirm config backup exists → exploit) rather than prose, mirroring the level-5 workflow anchor.
Clarify which referenced paths are sibling files in this skill's bundle versus external repo skills, so navigation targets are unambiguous and verifiable.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Lean, expert-level body that assumes competence — no padding explaining what an APT or LOTL is, and every section (kill-chain table, CONOPS, OPSEC, RoE gates, fidelity notes) earns its place. | 5 / 5 |
Actionability | Provides a concrete kill-chain table mapping phase→MITRE→action→executing skill, a copy-ready JSON threat-profile seed, and a numbered CONOPS to drop into conops.json; actual execution is delegated to other skills via explicit paths, leaving a minor gap versus fully copy-paste-ready code. | 4 / 5 |
Workflow Clarity | Clear phased sequence with explicit safety/abort gates ('EMERGENCY' abort on OT/safety systems, device-write authorization + config-backup check before exploiting) and deconfliction/cleanup recording; checkpoints are safety gates rather than technical validate→fix→retry feedback loops, leaving a minor gap. | 4 / 5 |
Progressive Disclosure | Well-organized into clearly labeled sections that point one level deep to external skill paths and an apt-groups reference; however those referenced paths (../../references/apt-groups.md, /skills/standard/...) are not part of this skill's bundle and cannot be verified, and some inline reference material could be split out. | 4 / 5 |
Total | 17 / 20 Passed |