rootline

分类: 数据与AI | 上传者: pablontiv | 下载: 0 | 版本: v1.0（最新）
解释、绘制图表、迁移、初始化、分析、应用。每当用户使用
更新日志: Source: GitHub https://github.com/pablontiv/rootline
目录结构

当前层级: tree/master/.claude/skills/rootline/
📄 ref-advanced.md 6.3 KB
📄 ref-query.md 4.7 KB
📄 ref-schema.md 3.8 KB
📄 ref-validate.md 3.9 KB
📄 SKILL.md 11.2 KB
SKILL.md

---
name: rootline
description: |
  Wrapper del CLI de rootline: validate, fix, set, describe, new, query, tree, stats,
  explain, graph, migrate, init, analyze, apply. Usar siempre que el usuario
  trabaje con documentos gestionados por rootline o schemas .stem — incluso sin
  decir "rootline". Usar este skill siempre que el usuario quiera validar
  documentos, verificar frontmatter, consultar registros estructurados, crear
  archivos scaffolded, bootstrap de consistencia, o inspeccionar schemas —
  incluso si solo dice "verificar si este archivo está bien", "que campos
  necesito", "buscar registros", "inferir schema", "hacer consistente",
  "set field", "cambiar estado", "actualizar campo", "mutar documento",
  "check if this file is valid", o "what fields does this need".
  (No para: descomposición en epics/stories = roadmap, investigación = hypothesize,
  sync de docs = update-docs, debugging de Go, operaciones YAML no-rootline.)
argument-hint: "[command] [args...]"
allowed-tools: Bash, Read, AskUserQuestion
---

# /rootline — CLI Operations

Rootline treats the filesystem as a database. This skill wraps all rootline CLI commands.

## Common Workflows

### Bootstrap Consistency (most common)

When a user has a folder with markdown files and wants rootline to make it consistent:

```bash
rootline init <path> --dry-run              # 1. Preview inferred schema
rootline init <path>                        # 2. Write .stem (add --force if exists)
rootline validate --all --output json       # 3. Check for errors
rootline fix --all --dry-run                # 4. Preview fixes (if errors found)
rootline fix --all                          # 5. Apply fixes
rootline validate --all                     # 6. Verify clean state
```

**Key considerations during bootstrap:**
- If `init` infers a field as `enum` but a parent `.stem` defines it as `string`, you'll get a `type-consistency` error. Resolution: remove the field from the local `.stem` and let it inherit from the parent.
- If `init` infers date-like fields as `enum`, change them to `type: string` — dates are open-ended, not a closed set.
- Always run `--dry-run` before writing to preview what will change.
- After init, show the user the effective schema (`rootline describe <path>`) so they see both local and inherited fields.

### Create → Validate Loop

When scaffolding a new document:

```bash
rootline describe <dir> --field schema.id.next   # Get next auto-ID
rootline new <filepath> --dry-run                 # Preview scaffold
rootline new <filepath>                           # Create document
rootline validate <filepath>                      # Verify it's valid
```

## Command Reference

| Command | Purpose | When to use |
|---------|---------|-------------|
| `validate` | Check documents against `.stem` schemas | User wants to verify files are correct |
| `fix` | Auto-repair validation errors | After validate finds fixable errors |
| `describe` | Show effective schema for a directory | User asks what fields are needed |
| `new` | Scaffold a document with frontmatter | User wants to create a new file |
| `query` | Search and filter records | User wants to find records by field values |
| `tree` | Hierarchical view with completion counts | User wants to see project structure |
| `stats` | Summary counts by type and state | User wants aggregate numbers |
| `explain` | Trace how a field got its value | User needs to debug derived/aggregated fields |
| `graph` | Dependency graph from wiki-links | User wants to visualize dependencies |
| `migrate` | Schema change detection and migration | User modified `.stem` and needs to adapt |
| `init` | Infer `.stem` from existing documents | User has docs but no schema yet |
| `analyze` | Run all inference detectors | User wants a full analysis report of patterns |
| `set` | Set field values (frontmatter or body sections) | User wants to mutate a document field with schema validation |
| `apply` | Apply inference results to .stem | User wants to update .stem based on analysis |

## Global Flags

All commands support:

- `--output json|table` (`-o`) — output format (default: `json`)
- `--field <dot-path>` — extract specific field from JSON output (repeatable)

Several transversal commands also support:

- `--where "expr"` — filter with expr-lang syntax (`==`, `!=`, `in`, `contains`, `exists`, `&&`)

## Routing

Based on user intent, apply the appropriate command group. Read the linked reference for detailed procedures, flags, and output formats.

### Validation (validate + fix)

Validate documents against `.stem` schemas and auto-fix errors.

```bash
rootline validate [file...] --output json          # Single or multiple files
rootline validate --all                             # All files in scope
rootline validate --all --where "estado == 'Pending'" --strict
rootline validate --staged                          # Git staging area only
rootline fix [file...] --dry-run                    # Preview fixes
rootline fix --all                                  # Fix all fixable errors
```

Key flags: `--all`, `--staged`, `--strict`, `--where`

**Procedure**: Determine target (file, directory, or all) → run validate → parse JSON → present errors with context → suggest `rootline fix` for fixable issues.

For detailed procedures and JSON output formats, see [ref-validate.md](ref-validate.md).

### Schema Inspection (describe + new)

Inspect effective schemas and scaffold new documents.

```bash
rootline describe <path> --output json              # Show merged schema
rootline describe <dir> --field schema.id.next      # Get next auto-ID
rootline describe <dir> --by-domain lifecycle_state # Filter by semantic domain
rootline new <filepath> --dry-run                   # Preview scaffold
rootline new <filepath>                             # Create document
```

Key flags: `--field` (extraction), `--by-domain` (filter by domain), `--dry-run`, `--force`

**Procedure**: For describe — show schema as table with field/type/required/values/source. For new — get next ID via describe, preview with dry-run, create, then validate.

For detailed procedures and JSON output formats, see [ref-schema.md](ref-schema.md).

### Queries & Exploration (query + tree + stats + explain)

Search, explore, and understand the document database.

```bash
rootline query --where "estado == 'Pending'" --limit 10
rootline query --count                              # Just the count
rootline tree --where "estado != 'Completed'"       # Filtered tree
rootline stats --from docs/epics/                   # Summary counts
rootline explain <file>                             # Trace field derivation
```

Key flags: `--where`, `--count`, `--limit`, `--from`

**Procedure**: Choose the right command based on what the user needs — query for filtering records, tree for structure, stats for aggregates, explain for debugging field values.

For detailed procedures and JSON output formats, see [ref-query.md](ref-query.md).

### Mutation (set)

Set field values on documents with schema validation. Supports frontmatter and body section fields.

```bash
rootline set file.md estado=listo-para-implementar       # Frontmatter enum
rootline set file.md bloqueador=""                        # Clear string
rootline set file.md investigacion=@findings.md           # Section from file
rootline set file.md investigacion+="New evidence"        # Append to section
rootline set --create file.md investigacion="GO"          # Create new section
rootline set --dry-run file.md estado=listo               # Preview changes
```

Key flags: `--dry-run` (preview), `--create` (create sections), `--no-validate` (skip post-validation)

**Procedure**: Determine target file + field=value pairs → schema lookup → pre-validate (enum check) → apply proposals → post-validate + rollback on failure. Section fields use the `heading` from `.stem` schema for exact match.

### Analysis (analyze)

Run all inference detectors and produce a structured report.

```bash
rootline analyze <directory> --output json          # Full analysis report
rootline analyze <directory> --output table          # Summary table
rootline analyze <directory> --incremental           # Only uncovered inferences
rootline analyze <directory> --field summary         # Extract summary only
```

Key flags: `--incremental` (filter inferences covered by .stem)

**Procedure**: Scans directory, extracts records, runs 16 detector categories — 13 data inference (field types, required fields, enums, constants, link types, back-refs, cross-refs, section patterns, invariants, sub-schemas, dependencies, traceability, structural rules) + 3 governance (domain coverage, schema coverage, validation gaps). Produces AnalyzeReport JSON with `version: 1`. With `--incremental`, filters out inferences already covered by the existing `.stem` schema. Governance detectors flag DDL-level gaps: fields without `domain`, directories without `.stem`, and fields with insufficient validation rules.

### Apply (apply)

Apply inference results to .stem files and document frontmatter.

```bash
rootline analyze <directory> --output json | rootline apply   # Pipe from analyze
rootline apply report.json                                    # From file
rootline apply report.json --output table                     # Human-readable
rootline apply report.json --dry-run                          # Preview changes
```

Key flags: `--dry-run` (show changes without applying)

**Procedure**: Reads an analyze report. Pre-phase: scaffolds `.stem` files for `missing_schema` inferences (creates minimal version-2 schema from observed frontmatter). Then applies schema-modifying inferences (extend_enum, add_required, add_default, set_type, untyped_field, sequence_incomplete) to the closest `.stem` file and data corrections (migrate_value, correct_value, add_field) to document frontmatter. Inferences with `requires_agent: true` are skipped with warnings.

### Advanced Operations (graph + migrate + init)

Dependency graphs, schema migrations, and schema inference.

```bash
rootline graph --format mermaid                     # Dependency visualization
rootline graph --check                              # Validate links (cycles + broken)
rootline migrate --dry-run                          # Detect schema changes
rootline migrate --rename old=new                   # Rename fields
rootline migrate --split                            # Flat → hierarchical .stem
rootline init --dry-run                             # Infer schema from docs
```

These commands are used less frequently. For detailed procedures, see [ref-advanced.md](ref-advanced.md).

## Output Handling

1. Always use `--output json` when parsing results programmatically
2. All JSON responses include `"version": 1` and `"kind": "rootline/<command>"`
3. Use `--output table` when displaying results directly to the user
4. Use `--field` for extracting specific values (e.g., `--field schema.id.next`)
5. Use `just validate` / `just fix-docs` as shortcuts for batch operations

## MCP Server

Rootline exposes all tools via MCP (Model Context Protocol). Two transport modes:

- **HTTP** (default): `rootline serve --addr 127.0.0.1:9200` — multi-consumer, stateless Streamable HTTP
- **Stdio** (legacy): `rootline serve --stdio` — single consumer, for Claude Code MCP config

HTTP mode serves `/mcp` (MCP Streamable HTTP) and `/health` (JSON status). Health tool (#11) is also available via MCP protocol.
登录后下载/点赞/收藏 ❤ 5 | ★ 0
rootline

目录结构

SKILL.md

举报内容

提示
