dgf1988/codewhale

Files

T

Hunter Bown cbcf35c1bd release: prepare v0.3.32 — finance tool, header redesign, expanded tests

Add Yahoo Finance quote tool with chart fallback, redesign header widget
with proportional truncation and context bar, refactor footer status strip,
expand test suite to 680+ tests, and fix blocking issues (usize underflow
in header, tempdir leak in finance tests, per-call HTTP client creation).

2026-04-11 18:58:56 -05:00

1.9 KiB

Raw Blame History

Project Instructions

This file provides context for AI assistants working on this project.

Project Type: Rust

Commands

Build: cargo build
Test: cargo test --workspace --all-features
Lint: cargo clippy --workspace --all-targets --all-features
Format: cargo fmt --all
Run: cargo run -p deepseek-tui

Documentation

See README.md for project overview, docs/ARCHITECTURE.md for internals.

Trimtab Workflow

This repo uses the Trimtab closed-loop protocol for self-verifying agentic development.

Protocol: .trimtab/init-trimtab-protocol.md (canonical — read this first)
Task graph: DEPENDENCY_GRAPH.md (crate deps + task deps with ready queue)
Task queue: AI_HANDOFF.md (6 open items with priorities)
Claude entrypoint: .claude/commands/init-trimtab.md
Codex skill: .codex/skills/init-trimtab/SKILL.md

No-self-verdict rule: The agent that wrote code must not be the one to declare it passes. Always use an independent verifier (fresh context or separate sub-agent).

DeepSeek-Specific Notes

Thinking Tokens: DeepSeek models output thinking blocks (ContentBlock::Thinking) before final answers. The TUI streams and displays these with visual distinction.
Reasoning Models: deepseek-reasoner and deepseek-r1 excel at step-by-step problem solving.
Large Context Window: 128k tokens. Use search tools to navigate efficiently.
API: OpenAI-compatible with Responses API preferred, chat completions as fallback. Base URL configurable for global (api.deepseek.com) or China (api.deepseeki.com).

Important Notes

Token/cost tracking inaccuracies: Token counting and cost estimation may be inflated due to thinking token accounting bugs. Use /compact to manage context, and treat cost estimates as approximate.
Modes: Three modes — Plan (read-only investigation), Agent (tool use with approval), YOLO (auto-approved). See docs/MODES.md for details.

1.9 KiB Raw Blame History