feat(personas): implement agent persona system with comprehensive infra #120
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
8f3ba70 to
ee34677
Compare
1. Needed comprehensive agent persona system for specialized AI workflows
2. Required isolated testing infrastructure and improved developer experience
3. improved ee2e testing framework
- **4 specialized agent personas**: Developer, Integration Tester, Merger, Rebaser
- **Comprehensive instructions**: Each persona has detailed workflows and rules
- **Task management**: 20+ item TODO list requirement for complex tasks
- **Quality gates**: 1500-line minimum read rule prevents partial understanding
- no more running getting-started.md manaully
- no more asing claude to run through it manually
- just sweet smooth e2e tests that run against an almost-real cluster
- **Isolated clusters**: Dynamic port allocation for branch-specific Kind clusters
- **Testing separation**: test-unit vs test-e2e with proper envtest setup
- **Developer tools**: .envrc, port detection, automated setup/teardown
- **Documentation**: Comprehensive agent instructions and workflow guides
- Created hack/agent-*.md files with specialized personas
- Implemented CLAUDE.md with mandatory persona selection
- Added comprehensive TODO list management patterns
- Established "delete more than you add" complexity principle
- Enhanced Makefiles with isolated cluster support
- Added dynamic port allocation via hack/find_free_port.sh
- Implemented branch-specific kubeconfig management
- Created comprehensive test framework with envtest
- added acp/test/e2e/ file structure
acp/test/e2e
├── framework
│ ├── basic_test.go
│ ├── framework.go
│ ├── getting_started
│ │ ├── suite_test.go
│ │ ├── test_getting_started.go
│ │ └── test_mcp_transport.go
│ └── suite_test.go
├── framework.go
└── getting_started
├── suite_test.go
└── test_getting_started.go
4 directories, 9 files
- No Headers/SessionID references remain in codebase: `grep -r "headers\|sessionId" acp/`
- Transport enum restricted: Check acp/api/v1alpha1/mcpserver_types.go:12
- CRDs regenerated properly: `kubectl get crd mcpservers.acp.humanlayer.dev -o yaml`
- Agent personas available: `ls hack/agent-*.md`
- CLAUDE.md enforces persona selection
- TODO list patterns implemented throughout instructions
- Isolated clusters: `make setup` creates branch-specific cluster
- Tests work: `make -C acp test-unit && make -C acp test-e2e`
- Build succeeds: `make -C acp build`
Major release introducing specialized agent personas, critical MCP security fixes,
and comprehensive infrastructure improvements. BREAKING: MCP Headers/SessionID
fields removed for security. Only stdio/http transports now supported.
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What problem(s) was I solving?
The Agent Control Plane project lacked specialized agent workflows for different engineering tasks, worker isolation for parallel development, and a robust testing framework for integration validation. This was causing workflow inefficiencies, test conflicts, and difficulty managing complex multi-agent development scenarios.
What user-facing changes did I ship?
How I implemented it
How to verify it
hack/launch_coding_workers.sh <branch> <persona>and verify specialized workflowsmake -C acp test-e2eto execute envtest-based integration testsmake -C acp fmt vet lint testto verify all quality checks passDescription for the changelog
Agent persona system: Specialized agent workflows, worker isolation, comprehensive e2e testing framework, and dynamic cluster management for reliable multi-agent development workflows.
🤖 Generated with Claude Code