hk-arch-audit
starsforks
How to install hk-arch-audit?
deepklarity/harness-kit/hk-arch-audit
$npx -y skills add deepklarity/harness-kit --skill hk-arch-auditRun comprehensive agent-native architecture review with scored principles. Audits a codebase against 8 agent-native architecture principles (Action Parity, Tools as Primitives, Context Injection, Shared Workspace, CRUD Completeness, UI Integration, Capability Discovery, Prompt-Na
SKILL.md
| 1 | # /hk-arch-audit — Agent-Native Architecture Review |
| 2 | |
| 3 | Conducts a comprehensive review of the codebase against agent-native architecture principles, launching parallel sub-agents for each principle and producing a scored report. |
| 4 | |
| 5 | ## Context |
| 6 | |
| 7 | <audit_context> $ARGUMENTS </audit_context> |
| 8 | |
| 9 | ## Core Principles |
| 10 | |
| 11 | | # | Principle | One-liner | |
| 12 | |---|-----------|-----------| |
| 13 | | 1 | Action Parity | Whatever the user can do, the agent can do | |
| 14 | | 2 | Tools as Primitives | Tools provide capability, not behavior | |
| 15 | | 3 | Context Injection | System prompt includes dynamic context about app state | |
| 16 | | 4 | Shared Workspace | Agent and user work in the same data space | |
| 17 | | 5 | CRUD Completeness | Every entity has full Create, Read, Update, Delete | |
| 18 | | 6 | UI Integration | Agent actions immediately reflected in UI | |
| 19 | | 7 | Capability Discovery | Users can discover what the agent can do | |
| 20 | | 8 | Prompt-Native Features | Features are prompts defining outcomes, not code | |
| 21 | |
| 22 | ## Workflow |
| 23 | |
| 24 | ### Step 1: Load reference material |
| 25 | |
| 26 | Read `references/agent-prompts.md` in this skill's directory to get the detailed prompt for each sub-agent. That file contains the exact instructions, search strategies, and output format for all 8 principle audits. |
| 27 | |
| 28 | ### Step 2: Check for single-principle audit |
| 29 | |
| 30 | If `$ARGUMENTS` specifies a single principle, only run that one sub-agent. Valid argument forms: |
| 31 | |
| 32 | | Argument | Principle | |
| 33 | |----------|-----------| |
| 34 | | `action parity` or `1` | Action Parity | |
| 35 | | `tools` or `primitives` or `2` | Tools as Primitives | |
| 36 | | `context` or `injection` or `3` | Context Injection | |
| 37 | | `shared` or `workspace` or `4` | Shared Workspace | |
| 38 | | `crud` or `5` | CRUD Completeness | |
| 39 | | `ui` or `integration` or `6` | UI Integration | |
| 40 | | `discovery` or `7` | Capability Discovery | |
| 41 | | `prompt` or `features` or `8` | Prompt-Native Features | |
| 42 | |
| 43 | For single-principle audits, run one Explore sub-agent and present its findings directly — no summary table needed. |
| 44 | |
| 45 | ### Step 3: Launch parallel sub-agents |
| 46 | |
| 47 | Launch 8 parallel sub-agents using the Task tool with `subagent_type: Explore`, one per principle. Each agent should: |
| 48 | |
| 49 | 1. Enumerate all relevant instances in the codebase (user actions, tools, contexts, data stores, etc.) |
| 50 | 2. Check compliance against the principle |
| 51 | 3. Provide a specific score in `X out of Y (percentage%)` format |
| 52 | 4. List specific gaps and recommendations |
| 53 | |
| 54 | Read `references/agent-prompts.md` for the exact prompt to give each agent. Pass the full agent prompt from that file — don't summarize it. |
| 55 | |
| 56 | ### Step 4: Compile summary report |
| 57 | |
| 58 | After all agents complete, compile the summary using this template: |
| 59 | |
| 60 | ```markdown |
| 61 | ## Agent-Native Architecture Review: [Project Name] |
| 62 | |
| 63 | ### Overall Score Summary |
| 64 | |
| 65 | | Core Principle | Score | Percentage | Status | |
| 66 | |----------------|-------|------------|--------| |
| 67 | | Action Parity | X/Y | Z% | STATUS | |
| 68 | | Tools as Primitives | X/Y | Z% | STATUS | |
| 69 | | Context Injection | X/Y | Z% | STATUS | |
| 70 | | Shared Workspace | X/Y | Z% | STATUS | |
| 71 | | CRUD Completeness | X/Y | Z% | STATUS | |
| 72 | | UI Integration | X/Y | Z% | STATUS | |
| 73 | | Capability Discovery | X/Y | Z% | STATUS | |
| 74 | | Prompt-Native Features | X/Y | Z% | STATUS | |
| 75 | |
| 76 | **Overall Agent-Native Score: X%** |
| 77 | |
| 78 | ### Status Legend |
| 79 | - Excellent (80%+) |
| 80 | - Partial (50-79%) |
| 81 | - Needs Work (<50%) |
| 82 | |
| 83 | ### Top 10 Recommendations by Impact |
| 84 | |
| 85 | | Priority | Action | Principle | Effort | |
| 86 | |----------|--------|-----------|--------| |
| 87 | |
| 88 | ### What's Working Well |
| 89 | |
| 90 | [List top 5 strengths] |
| 91 | ``` |
| 92 | |
| 93 | ## Success criteria |
| 94 | |
| 95 | - All 8 sub-agents complete their audits (or the single requested one) |
| 96 | - Each principle has a specific numeric score (X/Y format) |
| 97 | - Summary table shows all scores and status indicators |
| 98 | - Top 10 recommendations are prioritized by impact |
| 99 | - Report identifies both strengths and gaps |
People who liked this also viewed
6 picksType
- find-skillsHelps users discover and install agent skills when they ask questions like "how do I do X", "find a skill for X", "is there a skill that can...", or express interest in extending capabilities. This skSkillsJul 2026
- grill-meA relentless interview to sharpen a plan or design.SkillsJul 2026
- vercel-react-best-practicesVercel's official collection of agent skillsSkillsJul 2026
- improve-codebase-architectureScan a codebase for deepening opportunities, present them as a visual HTML report, then grill through whichever one you pick.SkillsJul 2026
- setup-matt-pocock-skillsConfigure this repo for the engineering skills — set up its issue tracker, triage label vocabulary, and domain doc layout. Run once before first use of the other engineering skills.SkillsJul 2026
- triageMove issues and external PRs through a state machine of triage roles — categorise, verify, grill if needed, and write agent-ready briefs.SkillsJul 2026
