name: agent-ops-build description: "Language-aware build orchestration that detects project language and runs appropriate build pipeline" category: extended invokes: [agent-ops-baseline, agent-ops-state] invoked_by: [agent-ops-validation] state_files: read: [constitution.md, baseline.md] write: [baseline.md, focus.md]

agent-ops-build

Language-aware build orchestration skill that detects project language and runs the appropriate build pipeline.

Purpose

Provide a unified build command that:

1. Auto-detects project language(s) from configuration files
2. Runs language-appropriate build stages
3. Reports findings in a consistent format
4. Integrates with agent-ops-baseline for comparison

When to Use

• Running a full build pipeline
• Validating changes before commit
• Capturing baseline metrics
• When constitution lacks explicit build commands

Detection Logic

flowchart TD
    A[Start] --> B{pyproject.toml?}
    B -->|Yes| C[Python Project]
    B -->|No| D{setup.py?}
    D -->|Yes| C
    D -->|No| E{package.json?}
    E -->|Yes| F[Node.js Project]
    E -->|No| G{*.sln?}
    G -->|Yes| H[.NET Project]
    G -->|No| I{*.csproj?}
    I -->|Yes| H
    I -->|No| J[Unknown - Check Constitution]

Language Detection

For mixed projects, run pipelines in priority order.

Build Pipelines

Python Pipeline

Stage 1: Dependencies  → uv sync / pip install
Stage 2: Format        → ruff format --check
Stage 3: Lint          → ruff check
Stage 4: Type Check    → mypy src/
Stage 5: Test          → pytest --cov
Stage 6: Coverage      → Check threshold

Commands:

uv sync
uv run ruff format --check .
uv run ruff check .
uv run mypy src/
uv run pytest --cov=src --cov-report=term-missing

TypeScript/Node.js Pipeline

Stage 1: Dependencies  → npm ci / pnpm install
Stage 2: Format        → prettier --check
Stage 3: Lint          → eslint
Stage 4: Type Check    → tsc --noEmit
Stage 5: Test          → vitest run / jest
Stage 6: Coverage      → Check threshold

Commands:

pnpm install
pnpm run format:check
pnpm run lint
pnpm run typecheck
pnpm run test:coverage

C#/.NET Pipeline

Stage 1: Dependencies  → dotnet restore
Stage 2: Format        → dotnet format --verify-no-changes
Stage 3: Build         → dotnet build -c Release
Stage 4: Test          → dotnet test --collect:"XPlat Code Coverage"
Stage 5: Coverage      → ReportGenerator (if available)

Commands:

dotnet restore
dotnet format --verify-no-changes
dotnet build -c Release --no-restore
dotnet test -c Release --no-build --collect:"XPlat Code Coverage"

Workflow

1. Detection Phase

Detecting project type...
  ✓ Found pyproject.toml → Python
  ✓ Found package.json → Node.js
  
Detected: Python + Node.js (mixed project)

2. Configuration Phase

Check for existing configuration:

• Constitution commands (highest priority)
• Package manager detection (uv vs pip, pnpm vs npm)
• Tool configuration files (ruff.toml, eslint.config.js)

3. Execution Phase

Run each stage, stopping on first failure:

Python Build Pipeline
═════════════════════

[1/6] Dependencies
  $ uv sync
  ✓ Resolved 45 packages (2.1s)

[2/6] Format Check
  $ uv run ruff format --check .
  ✓ 23 files checked (0.3s)

[3/6] Lint
  $ uv run ruff check .
  ✓ 0 issues (0.4s)

[4/6] Type Check
  $ uv run mypy src/
  ✓ Success, no errors (1.2s)

[5/6] Tests
  $ uv run pytest --cov=src
  ✓ 45 passed (3.1s)

[6/6] Coverage
  ✓ 84.2% (threshold: 80%)

═════════════════════
✓ BUILD SUCCESSFUL
  Total time: 7.1s

4. Report Phase

Generate consistent output:

## Build Report

**Project**: my-project
**Languages**: Python, Node.js
**Date**: 2024-01-15 10:30:00

### Python

| Stage | Status | Time | Notes |
|-------|--------|------|-------|
| Dependencies | ✓ | 2.1s | 45 packages |
| Format | ✓ | 0.3s | 23 files |
| Lint | ✓ | 0.4s | 0 issues |
| Type Check | ✓ | 1.2s | - |
| Tests | ✓ | 3.1s | 45 passed |
| Coverage | ✓ | - | 84.2% |

### Node.js

| Stage | Status | Time | Notes |
|-------|--------|------|-------|
| Dependencies | ✓ | 4.2s | 120 packages |
| Format | ✓ | 0.5s | - |
| Lint | ✓ | 1.1s | 0 warnings |
| Type Check | ✓ | 2.3s | - |
| Tests | ✓ | 5.2s | 78 passed |
| Coverage | ✓ | - | 72.1% |

### Summary

- **Status**: ✓ PASSED
- **Total Time**: 19.4s
- **Warnings**: 0
- **Errors**: 0

Constitution Integration

If constitution.md has explicit commands, prefer those:

# Constitution

## Commands
- **Build**: `make build`
- **Test**: `make test`
- **Lint**: `make lint`

The skill will use constitution commands over auto-detected ones.

Baseline Integration

When --baseline flag is used or baseline.md exists:

1. Run build pipeline
2. Compare results to baseline
3. Report differences

Comparing to baseline (2024-01-14)...

| Metric | Baseline | Current | Change |
|--------|----------|---------|--------|
| Tests | 45 | 48 | +3 |
| Coverage | 82.1% | 84.2% | +2.1% |
| Lint | 0 | 0 | - |
| Type Errors | 0 | 0 | - |

✓ No regressions detected

Options

Error Handling

Stage Failure

[3/6] Lint
  $ uv run ruff check .
  ✗ 3 issues found

  src/auth.py:45:1: E501 Line too long (120 > 100)
  src/users.py:12:5: F401 Unused import 'os'
  src/users.py:30:1: E302 Expected 2 blank lines

  Hint: Run with --fix to auto-fix 2 issues
  
═════════════════════
✗ BUILD FAILED at stage: Lint

Missing Tools

[2/6] Format Check
  ⚠ ruff not found

  Hint: Install with 'uv add --dev ruff'
  Skipping format check...

Examples

Basic Build

Run the build pipeline

Specific Language

Run Python build only

With Auto-fix

Run build and fix any auto-fixable issues

Compare to Baseline

Run build and compare to baseline

Output Files

The skill can optionally output:

• build-report.md — Markdown report
• build-report.json — Machine-readable JSON
• Updates to baseline.md — If capturing new baseline

Integration with Other Skills

Troubleshooting

"No project detected"

Ensure you have one of:

• pyproject.toml / setup.py (Python)
• package.json (Node.js)
• *.sln / *.csproj (.NET)

Or add explicit commands to constitution.md.

"Tool not found"

Install required tools:

• Python: uv, ruff, mypy, pytest
• Node.js: pnpm/npm, eslint, prettier
• .NET: dotnet CLI

"Coverage below threshold"

Either:

1. Add more tests to increase coverage
2. Adjust threshold in configuration
3. Use --coverage-threshold to override