charieswei
/
TradingAgents
mirror of https://github.com/TauricResearch/TradingAgents.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity
259 Commits 2 Branches 6 Tags 543 MiB
Commit Graph

7 Commits

Author SHA1 Message Date
Ahmet Guzererler da7f078b5b removed hooks 2026-03-20 12:58:21 +01:00
ahmet guzererler d2e967c4fc
simplify Architect agent prompt from ~165 to ~45 lines (#27) 
Condense 7 verbose phases into 6 concise steps, remove redundant
error-handling table and bloated principles section while preserving
all critical rules (max 3 context calls, model: sonnet, self-contained
prompts, parallel spawning).

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
 2026-03-19 14:47:34 +01:00
Ahmet Guzererler 659cb3bafb architext change 2026-03-19 14:27:41 +01:00
ahmet guzererler d2af8991ed
fix: streamline Architect agent to stop stalling on context gathering (#25) 
Collapsed 9 sequential phases into 5 core + 2 optional phases. The agent
was spending all its turns reading memory files, ADRs, and conventions
before ever spawning implementation agents. Now it caps initial context
gathering at 3 tool calls, skips memory-reader skill invocation (CLAUDE.md
already has conventions), writes plans as plain text instead of using
EnterPlanMode, and makes cleanup/memory-update phases optional.

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
 2026-03-19 13:46:16 +01:00
Ahmet Guzererler 97ab49bb99 skills 2026-03-19 10:32:29 +01:00
ahmet guzererler 8fcd58ad0f
feat: memory system v2 — skills, context files, post-commit hook (#21) 
* gitignore

* feat: memory system v2 — builder/reader skills, context files, post-commit hook

Replace architecture-coordinator skill with two focused skills:
- memory-builder: extracts repo knowledge into 4-layer memory system with
  mandatory verification steps (deps, model names, agent count, ADR cross-refs)
- memory-reader: loads memory at session start, enforces ADR constraints,
  detects staleness

Generate 5 verified context files under docs/agent/context/:
- ARCHITECTURE.md (105 lines) — system design, pipelines, LLM tiers
- COMPONENTS.md (188 lines) — 17 agent factories, extension guides, test inventory
- CONVENTIONS.md (88 lines) — coding rules with source citations
- TECH_STACK.md (75 lines) — 22 deps from pyproject.toml with versions
- GLOSSARY.md (92 lines) — 50+ terms across 9 domains

Add PostToolUse hook in .claude/settings.json to remind agent to update
CURRENT_STATE.md after git commits.

All 30 factual claims audited and verified against source code.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
 2026-03-18 18:38:03 +01:00
Ahmet Guzererler 7f22b8e889 feat: add architecture-coordinator skill for mandatory ADR reading protocol 
New Claude Code skill that enforces reading docs/agent/CURRENT_STATE.md,
decisions/, and plans/ before any code changes. Includes conflict resolution
protocol that stops work and quotes the violated ADR rule when user requests
conflict with established architectural decisions.

Files:
- .claude/skills/architecture-coordinator/SKILL.md
- .claude/skills/architecture-coordinator/references/adr-template.md
- .claude/skills/architecture-coordinator/references/reading-checklist.md

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
 2026-03-17 20:26:15 +01:00