5b0d00a799
feat(config): T3 backend/frontend → senior devs; update agency-agents submodule
...
T3 squad leads now use dedicated senior roles instead of sharing
architects with T2:
- T3 backend: backend-architect → senior-backend-developer
- T3 frontend: frontend-architect → senior-frontend-developer
Clean tier separation:
T2 = architects (design)
T3 = senior devs (lead + implement-or-delegate)
T4 = developers (pure implementation)
Depends on: coding-with-hans-heinemann/agency-agents#2
2026-03-16 11:31:32 -04:00
461b36bc5d
fix(config): T2 backend → backend-architect (per review)
2026-03-16 11:13:19 -04:00
9c1794c58a
fix(config): T3 backend → backend-architect; update agency-agents submodule
...
- T3 backend squad lead should be an architect, not a developer (per review)
- Submodule updated to include frontend-architect + backend-developer roles
2026-03-16 10:36:33 -04:00
8adab6fbc5
refactor(config): replace senior-developer with role-specific specialists
...
- T2 frontend: software-architect → frontend-architect
- T3 backend: senior-developer → backend-developer
- T3 frontend: senior-developer → frontend-architect
- T3 default: senior-developer → backend-developer
- T4 backend: backend-architect → backend-developer (implementation, not architecture)
- T4 default: senior-developer → backend-developer
Depends on: coding-with-hans-heinemann/agency-agents#1 (new agent roles)
2026-03-16 10:11:47 -04:00
d02faf5cac
feat(config): expand role_registry + fix T4 default runtime
...
Role registry additions:
- T2: add ai, security, mobile domains
- T3: add data, ai, security, mobile, database, devops, docs domains
- T4: add data (data-engineer), embedded (firmware-engineer)
- T5: add accessibility, e2e, frontend, data verifier roles
TeamRunner consistency fix:
- T4 briefs now default to 'coding_agent' runtime (Claude Code)
per build spec: 'Claude Code as default T4 runtime'
T3 can still override preferred_runtime per task
2026-03-16 09:14:22 -04:00
aef553bdc8
refactor: T4 fast-cheap; move adapter registries from code to team.yaml
2026-03-16 01:14:35 -04:00
7b1cf7315c
refactor(team_runner): make runtimes config-driven — replace hardcoded slots with dict
...
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com >
2026-03-16 00:50:17 -04:00
bd96a83069
fix: derive LLM provider from adapter, not config
...
Remove redundant models.provider from team.yaml. Each adapter knows its
own provider key — AnthropicAdapter always looks up 'anthropic' in the
capability_map. This avoids a footgun where adapters.llm and models.provider
could disagree.
Future adapters (OpenAIAdapter, OllamaAdapter) will hardcode their own key
the same way.
2026-03-15 23:47:52 -04:00
8524b63a76
fix: read default_temperature from team.yaml; update docstrings
...
- Add default_temperature: 0 to config/team.yaml models block
- Read self._default_temperature from models cfg in __init__
- Use self._default_temperature as fallback in complete() instead of hardcoded 0
- Update class docstring to document both default_max_tokens and default_temperature
- Update complete() context param docs to reference team.yaml keys
2026-03-15 21:40:05 -04:00
6856f10c27
fix(adapter/llm): make max_tokens configurable via team.yaml models.default_max_tokens
2026-03-15 18:55:57 -04:00
eaf7fd8f6f
feat: initial bootstrap — structure, task_brief, blackboard, adapter bases, escalation, prompts
...
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com >
2026-03-15 02:19:14 -04:00
