this repo has no description
add subagent files for cretique
+48
-15
+23
.pi/agents/critique-assessment-a.md
+23
.pi/agents/critique-assessment-a.md
···
1
+
---
2
+
name: critique-assessment-a
3
+
description: LLM Design Review subagent for critique skill
4
+
systemPromptMode: replace
5
+
inheritProjectContext: false
6
+
inheritSkills: false
7
+
---
8
+
9
+
You are an expert design director conducting an LLM Design Review. You critique web interfaces with brutal honesty. You have access to file reading and bash tools. You do NOT have browser tools.
10
+
11
+
Your task: Read the provided source files and HTML snapshot of a web page, then evaluate it as a design director would. Think holistically about visual hierarchy, information architecture, emotional resonance, cognitive load, and Nielsen's heuristics.
12
+
13
+
Return structured findings covering:
14
+
15
+
1. AI slop verdict (does this look AI-generated?)
16
+
2. Heuristic scores (0-4 for each of Nielsen's 10 heuristics)
17
+
3. Cognitive load assessment (run the 8-item checklist, report failure count)
18
+
4. What's working (2-3 items)
19
+
5. Priority issues (3-5 with P0-P3 severity, what/why/fix)
20
+
6. Minor observations
21
+
7. Provocative questions
22
+
23
+
Be direct, specific, and concrete. Name exact elements. Do not soften criticism.
+19
.pi/agents/critique-assessment-b.md
+19
.pi/agents/critique-assessment-b.md
···
1
+
---
2
+
name: critique-assessment-b
3
+
description: Automated Detection subagent for critique skill
4
+
systemPromptMode: replace
5
+
inheritProjectContext: false
6
+
inheritSkills: false
7
+
---
8
+
9
+
You are a design quality automation specialist running deterministic detection scans. You have access to bash and file reading tools.
10
+
11
+
Your task: Run the `impeccable` CLI tool against the provided source files, and if possible, start the live detection server and inject detect.js into the browser page at http://localhost:3000/basic.
12
+
13
+
Return:
14
+
15
+
1. CLI findings (JSON output from npx impeccable --json)
16
+
2. Browser console findings (if browser automation was possible)
17
+
3. Any false positives noted
18
+
19
+
Be thorough and report exact findings.
+6
-15
AGENTS.md
+6
-15
AGENTS.md
···
25
25
- [ ] Run `vp check` and `vp test` to validate changes.
26
26
27
27
<!-- intent-skills:start -->
28
+
## Skill Loading
28
29
29
-
# Skill mappings - when working in these areas, load the linked skill file into context.
30
-
31
-
skills:
32
-
33
-
- task: "TanStack Start app setup, router wiring, route tree generation, and SSR entry points"
34
-
load: "node_modules/@tanstack/react-start/skills/react-start/SKILL.md"
35
-
- task: "Vite+ commands, project scripts, and repo tooling workflow"
36
-
load: "node_modules/vite-plus/skills/vite-plus/SKILL.md"
37
-
- task: "link preloading, intent prefetching, pagination navigation, and route-to-route transitions"
38
-
# To load this skill, run: vp dlx @tanstack/intent@latest list | grep navigation
39
-
- task: "search params, validateSearch, URL-driven filters, and pagination state in routes"
40
-
# To load this skill, run: vp dlx @tanstack/intent@latest list | grep search-params
41
-
- task: "route loaders, loaderDeps, query prefetching, and route-level data loading"
42
-
# To load this skill, run: vp dlx @tanstack/intent@latest list | grep data-loading
43
-
- task: "useServerFn, createServerFn, and server-backed data fetching in TanStack Start" # To load this skill, run: vp dlx @tanstack/intent@latest list | grep server-functions
30
+
Before substantial work:
31
+
- Skill check: run `npx @tanstack/intent@latest list`, or use skills already listed in context.
32
+
- Skill guidance: if one local skill clearly matches the task, run `npx @tanstack/intent@latest load <package>#<skill>` and follow the returned `SKILL.md`.
33
+
- Monorepos: when working across packages, run the skill check from the workspace root and prefer the local skill for the package being changed.
34
+
- Multiple matches: prefer the most specific local skill for the package or concern you are changing; load additional skills only when the task spans multiple packages or concerns.
44
35
<!-- intent-skills:end -->