trace-claude-code
Automatically trace Claude Code conversations to Braintrust for observability. Captures sessions, conversation turns, and tool calls as hierarchical traces.
Release preparation workflow - security audit → E2E tests → review → changelog → docs
$ npx -y skills add parcadei/Continuous-Claude-v3 --skill release --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/releaseContext preview
The summary Claude sees to decide when to auto-load this skill.
Release preparation workflow - security audit → E2E tests → review → changelog → docs
name: release description: Release preparation workflow - security audit → E2E tests → review → changelog → docs
Structured release preparation to ship with confidence.
┌─────────┐ ┌─────────┐ ┌──────────────┐ ┌──────────┐ ┌─────────┐ │ aegis │───▶│ atlas │───▶│ review-agent │───▶│ herald │───▶│ scribe │ │ │ │ │ │ │ │ │ │ │ └─────────┘ └─────────┘ └──────────────┘ └──────────┘ └─────────┘ Security E2E Final Version Release audit tests review bump notes
| # | Agent | Role | Output | |---|-------|------|--------| | 1 | **aegis** | Security vulnerability scan | Security report | | 2 | **atlas** | Run full E2E test suite | Test report | | 3 | **review-agent** | Final release review | Release approval | | 4 | **herald** | Version bump, changelog generation | Updated version files | | 5 | **scribe** | Release notes, documentation | RELEASE.md, docs |
1. **Security first**: Catch vulnerabilities before they ship 2. **E2E tests**: Verify full system works end-to-end 3. **Final review**: Human-in-the-loop approval 4. **Version bump**: Only after approval 5. **Documentation**: Ship with proper release notes
Task( subagent_type="aegis", prompt=""" Security audit for release: [VERSION] Scan for: - Dependency vulnerabilities (npm audit, pip audit) - Hardcoded secrets/credentials - SQL injection, XSS, CSRF risks - Authentication/authorization issues - Insecure configurations Output: Security report with severity levels """ )
Task( subagent_type="atlas", prompt=""" Run E2E tests for release: [VERSION] Execute: - Full E2E test suite - Critical path tests - Integration tests - Performance benchmarks (if applicable) Output: Test report with pass/fail counts """ )
Task( subagent_type="review-agent", prompt=""" Final release review: [VERSION] Review: - Security audit results - E2E test results - Changes since last release (git log) - Breaking changes - Migration requirements Output: RELEASE_APPROVED or RELEASE_BLOCKED with reasons """ )
Task( subagent_type="herald", prompt=""" Prepare release: [VERSION] Tasks: - Bump version in package.json/pyproject.toml - Generate CHANGELOG.md entry - Update version constants in code - Tag commit (don't push yet) Follow semantic versioning. """ )
Task( subagent_type="scribe", prompt=""" Write release notes: [VERSION] Include: - Summary of changes - New features - Bug fixes - Breaking changes - Migration guide (if needed) - Contributors Output: RELEASE.md or update docs """ )
/release --major → Full workflow with migration guide
/release --minor → Full workflow, lighter security review
/release --patch → Security + tests + quick review
/release --hotfix → Expedited: aegis → atlas → herald
User: /release v2.0.0 Claude: Starting /release workflow for v2.0.0... Phase 1: Security audit... [Spawns aegis] ✅ No critical vulnerabilities ⚠️ 2 low-severity issues (documented) Phase 2: E2E tests... [Spawns atlas] ✅ 156/156 E2E tests passing Phase 3: Final review... [Spawns review-agent] ✅ RELEASE_APPROVED - 47 commits since v1.9.0 - 3 new features - 12 bug fixes - No breaking changes Phase 4: Version bump... [Spawns herald] ✅ Version bumped to 2.0.0 ✅ CHANGELOG.md updated ✅ Git tag created Phase 5: Release notes... [Spawns scribe] ✅ RELEASE-v2.0.0.md created ┌─────────────────────────────────────────┐ │ Release v2.0.0 Ready │ ├─────────────────────────────────────────┤ │ Security: ✅ Passed │ │ Tests: ✅ 156/156 │ │ Review: ✅ Approved │ │ │ │ Next steps: │ │ 1. git push origin v2.0.0 │ │ 2. Create GitHub release │ │ 3. Deploy to production │ └─────────────────────────────────────────┘
The workflow stops if:
Phase 1: Security audit... ❌ CRITICAL: SQL injection in user.py:45 Release blocked. Fix critical issues before proceeding.
A persistent, learning, multi-agent development environment built on Claude Code Continuous Claude transforms Claude Code into a continuously learning system that maintains context across sessions, orchestrates specialized agents, and eliminates wasting
Repo: parcadei/Continuous-Claude-v3
Automatically trace Claude Code conversations to Braintrust for observability. Captures sessions, conversation turns, and tool calls as hierarchical traces.
Guide for integrating Agentica SDK with Claude Code CLI proxy
Reference guide for Agentica multi-agent infrastructure APIs