Agent-Skills.md

Agent Skills: Skill: test-primitives

Test CAT LAB safe primitives including file locking, atomic writes, and task validation

UncategorizedID: Reneromero08/agent-governance-system/test-primitives

Repository

Reneromero08/agent-governance-system
Reneromero08License: NOASSERTION

Install this agent skill to your local

pnpm dlx add-skill https://github.com/Reneromero08/agent-governance-system/tree/HEAD/CAPABILITY/SKILLS/utilities/test-primitives

Skill Files

Browse the full folder contents for test-primitives.

Download Skill

Loading file tree…

CAPABILITY/SKILLS/utilities/test-primitives/SKILL.md

Skill Metadata

Name
test-primitives
Description
Test CAT LAB safe primitives including file locking, atomic writes, and task validation

required_canon_version: >=3.0.0

Skill: test-primitives

Version: 0.1.0

Status: Active

Purpose

Tests the safe primitives used throughout the AGS system, including:

  • Platform-specific file locking (Windows/Unix)
  • Atomic JSONL write and rewrite operations
  • Streaming JSONL reader
  • Task state transition validation
  • Task spec validation
  • File hashing (SHA-256)
  • Validator constants verification

Trigger

Use when:

  • Verifying that safe primitives work correctly on the current platform
  • After system updates that might affect file operations
  • During CI/CD pipeline validation
  • Testing cross-platform compatibility

Inputs

JSON object with optional fields:

  • run_tests (boolean, optional): Whether to run actual tests (default: true)
  • dry_run (boolean, optional): Return mock results for deterministic testing

Outputs

JSON object:

  • tests_run (integer): Number of tests executed
  • tests_passed (integer): Number of tests that passed
  • tests_failed (integer): Number of tests that failed
  • details (array): List of test results with pass/fail indicators
  • summary (string): Human-readable summary
  • validator_semver (string): Current validator semantic version
  • build_id (string): Validator build fingerprint
  • durable_roots_count (integer): Number of durable roots
  • catalytic_roots_count (integer): Number of catalytic roots

Constraints

  • Creates temporary files during test execution
  • All temporary files are cleaned up after tests
  • Platform-specific behavior for file locking
  • Non-destructive (does not modify repository files)

Tests Performed

  1. File locking (Windows/Unix) - Exclusive file lock acquire/release
  2. Atomic JSONL write - Append lines atomically
  3. Atomic JSONL rewrite - Transform and rewrite JSONL atomically
  4. Streaming JSONL reader - Read with pagination and filtering
  5. Task state transitions - Validate allowed state changes
  6. Task spec validation - Validate task specification format
  7. File hashing (SHA-256) - Compute file hashes
  8. Constants and build ID - Verify validator constants

Fixtures

  • fixtures/basic/ - Deterministic test with mock results

required_canon_version: >=3.0.0

Skill: test-primitives Skill | Agent Skills