Content
71%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.
The skill body is highly actionable, with copy-paste-ready commands and a coherent scan-relay-coerce workflow validated by a criteria checklist. Its main weaknesses are a lack of progressive disclosure — bundle reference files and the agent.py script are never referenced while their content is partially duplicated inline — and mild verbosity from concept re-explanation and redundant tables.
Suggestions
Add explicit one-level-deep pointers to the bundle files (e.g. 'Full flag reference: see references/api-reference.md; MITRE/NIST mapping: see references/standards.md; automation wrapper: scripts/agent.py') and remove the inlined 'Coercion Method Reference' and MITRE tables that duplicate them.
Embed inline validation checkpoints between steps — after Step 3 confirm the coerced authentication arrived at the relay listener, and after Step 5 verify dc.pfx authenticates before DCSync — instead of only a terminal checklist.
Trim the Overview's explanation of what forced authentication and NTLM relay are (Claude already knows these) down to the tool- and method-specific facts.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly command-dense and efficient, but includes unnecessary material: an Overview that re-explains forced-authentication/NTLM-relay concepts Claude already knows, and redundant tables ('Coercion Method Reference', 'Tools and Resources', MITRE table) whose content duplicates both the Overview text and references/api-reference.md and references/standards.md. Not score 2 because there is no fluffy library-comparison padding and the bulk is executable commands; not score 4 because the concept re-explanation and table duplication are trimmable overhead. | 3 / 5 |
Actionability | Every workflow step provides copy-paste-ready commands with real flags and placeholders explained inline: 'coercer scan -u ... -t 10.0.0.10 -l 10.0.0.50', 'certipy relay -target http://CA.CORP.LOCAL -template DomainController', 'impacket-ntlmrelayx -t http://CA.CORP.LOCAL/certsrv/certfnsh.asp -smb2support --adcs', plus install commands in Prerequisites. Not score 4 because the examples cover the common cases (scan, coerce, filter, fuzz, ESC8 relay, RBCD relay) completely and would run as written. | 5 / 5 |
Workflow Clarity | A clear 7-step sequence (scan → relay → coerce → use result → alternative chains) with a terminal 'Validation Criteria' checklist covering each objective. Not score 5 because validation lives only in an end-of-document checklist rather than inline checkpoints or feedback loops — e.g. no 'confirm the coerced auth hit the listener before running certipy auth' step and no recovery guidance when a coercion attempt fails. Not score 3 because the sequence is coherent, checkpoints are enumerated, and each step states its purpose and preconditions. | 4 / 5 |
Progressive Disclosure | The bundle provides references/api-reference.md (full flag reference), references/standards.md (MITRE/NIST mapping), and scripts/agent.py, but the body never points to any of them — no 'See references/api-reference.md' navigation exists. Instead it inlines duplicate content (method tables, command examples also present in api-reference.md, MITRE mapping duplicated in standards.md), leaving the reference files orphaned and the reader without navigation. Not score 2 because the body itself is well-sectioned and the inlined material is compact, but signaling references and moving the duplicated detail would lift it to 4. | 3 / 5 |
Total | 15 / 20 Passed |