haipipe-project-inspect

name	haipipe-project-inspect
description	Read-only specialist for haipipe-project. Reviews project structure, summarizes, generates inventory, prints overview. Touches no files. Called by /haipipe-project orchestrator. Direct invocation works for project audit work.
argument-hint	[function] [project_id_or_path] [args...]
allowed-tools	Read, Grep, Glob, Bash
metadata	{"version":"0.1.0","last_updated":"2026-05-31","summary":"Read-only specialist for haipipe-project.","changelog":["1.0.0 (2026-05-31): baseline metadata added."]}

Skill: haipipe-project-inspect

Read-only specialist. Reviews, summarizes, inventories, and overviews project state without modifying anything. Called by the /haipipe-project orchestrator; can also be invoked directly.

Function axis: review | summarize | inventory | overview | scan-status

---

Commands

/haipipe-project-inspect                                    -> overview of all examples/
/haipipe-project-inspect overview [project_id]              -> overview of one project (or all)
/haipipe-project-inspect review <project_id>                -> structural audit (group/task rules)
/haipipe-project-inspect summarize <project_id>             -> generate summary doc
/haipipe-project-inspect inventory [project_id]             -> file inventory (uses ref/inventory/*.py)
/haipipe-project-inspect scan-status <task_dir> [key] [txt] -> scan B01 eval task, update status.json, format txt

---

Dispatch Table

Invocation     Ref file(s)                              Function file
-------------- ---------------------------------------- ----------------
overview       ../haipipe-project/ref/project-          fn/overview.md
               structure.md
review         ../haipipe-project/ref/project-          fn/review.md
               structure.md +
               ../haipipe-project/ref/code-structure.md
summarize      ../haipipe-project/ref/project-          fn/summarize.md
               structure.md
inventory      ref/inventory/*.py                       fn/inventory.md
scan-status    ref/scan_status/scan_status.py +         fn/scan-status.md
               ref/scan_status/status_formatter.py +
               ref/scan_status/scan_groups_template.json

---

Step-by-Step Protocol

Step 0: Read the relevant umbrella ref(s) per the dispatch table.

Step 1: Parse args. function in { overview, review, summarize, inventory, scan-status, (none) } target = project_id, task_dir, or path; default = scan all of examples/

Step 2: Read this skill's relevant fn doc.

Step 3: For inventory, use the helper scripts under ref/inventory/: - helpers.py — building blocks - refresh_all.py — full repo scan - render.py — output formatter For scan-status, use the scripts under ref/scan_status/: - scan_status.py — scans sbatch/runs/results, writes status.json - status_formatter.py — reads status.json, writes human-readable txt - scan_groups_template.json — template for scan_groups.json config

Step 4: Execute the function. NO writes. NO mv. NO rm.

Step 5: Emit the structured tail:

status:    ok | blocked | failed
summary:   2-3 sentences on findings (violations? counts? doc generated?)
artifacts: [paths read, output paths if any (e.g. summary.md emitted)]
next:      suggested next command (often /haipipe-project-organize if violations
           found, or /haipipe-task for missing project)

---

Risk Profile

Read-only. allowed-tools deliberately excludes Write/Edit. If a write is needed (e.g. summarize wants to drop a summary.md), the skill calls Bash to redirect output to a file under the project's own folder — but never modifies code, configs, or other projects' state.

If a violation is found that needs fixing, hand off to /haipipe-project-organize. Don't fix here.