golem15/msd-core
4
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
256 Commits 4 Branches 2 Tags
6c59948c46fe5c8b4189e9c56bb697c1e54fba89
Commit Graph

8 Commits

Author SHA1 Message Date
Lex Christopherson
d30893a834 feat: pre-compute wave numbers at plan time 
Wave numbers now computed during plan-phase and stored in PLAN.md
frontmatter. Execute-phase reads wave directly instead of deriving
from depends_on at runtime.

- Add assign_waves step to plan-phase workflow
- Add wave field to frontmatter (plan-format, phase-prompt template)
- Simplify execute-phase: remove analyze_dependencies and group_into_waves
- Replace with group_by_wave that just reads frontmatter integers

Co-Authored-By: Claude <noreply@anthropic.com>
 2026-01-13 18:10:20 -06:00
Lex Christopherson
0dd4cedf5a perf(execute-phase): remove subagent-only context from orchestrator 
Orchestrator loads only what it needs:
- execute-phase.md workflow (orchestration)
- subagent-task-prompt.md (template to spawn agents)

Subagents load execution context themselves via template:
- execute-plan.md, summary.md, checkpoints.md, tdd.md

~80% context reduction for orchestrator (~2000 lines saved)

Co-Authored-By: Claude <noreply@anthropic.com>
 2026-01-13 17:26:18 -06:00
Lex Christopherson
46cf4b11de feat(execute-phase): add deviation rules, commit rules, and workflow references 
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
 2026-01-13 17:01:10 -06:00
Lex Christopherson
e30e387c00 refactor(execute-phase): wave-based blocking execution 
Replace background agents + polling with blocking Task calls.
Orchestrator stays lean (~15% context), subagents load full workflow.

- Archive old execute-phase command and workflow
- Add subagent-task-prompt.md template (orchestrator→subagent contract)
- Rewrite execute-phase with wave-based parallelization
- Dependency analysis groups plans into execution waves
- Parallel Task calls within waves, sequential across waves
- No TaskOutput polling - Task blocks until completion

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
 2026-01-13 16:06:17 -06:00
Lex Christopherson
b810d1ddca fix(execute-phase): add one-subagent-per-plan constraint 
Context isolation is the primary goal, not parallelization.
Even sequential plans spawn separate subagents for fresh 200k context.

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
 2026-01-12 14:34:37 -06:00
Lex Christopherson
31f565064d fix(10-02): sync execute-phase command with execute-plan content 
- Added SlashCommand to allowed-tools
- Added references/tdd.md to execution_context
- Added execution_strategies, deviation_rules, commit_rules sections
- Updated success_criteria to include per-task commits
- Now fully aligned with execute-plan.md command structure
 2026-01-12 13:24:49 -06:00
Lex Christopherson
8e8fba2793 fix(10-02): load execute-plan.md context in execute-phase command 
- Command now loads both workflows (execution + orchestration)
- Agents commit tasks normally (removed DON'T COMMIT instruction)
- Orchestrator only commits metadata after all agents complete
- No duplication - single source of truth for execution logic
 2026-01-12 13:23:16 -06:00
Lex Christopherson
18a1fd17c6 feat(10-03): create execute-phase command 
- New slash command for parallel phase execution
- References execute-phase.md workflow
- Requires phase number as argument
- Uses parallel agent spawning for concurrent plans
 2026-01-12 13:14:53 -06:00