Tianyi Cui 1ad4615fab docs(dsh-code-review): frame skill as guidance, emphasize independent broad reasoning
The skill read like an authoritative checklist; clearing its items could be
mistaken for a complete review. Reframe it explicitly as guidance — a
where-to-look map that lowers startup cost, not a definition of a sufficient
review. Add a "How to think about a review" section drawing on the patterns
from the /code-review, /review, requesting-code-review, and
receiving-code-review skills: reason from the code independently, sweep all
aspects (correctness, concurrency, security, design/approach, tests, docs,
…), verify before flagging, calibrate confidence and suppress noise, and lead
with severity. Add a receiving-review note so authors evaluate findings on
merit rather than following them blindly.
2026-06-18 00:56:06 +08:00
2026-06-16 14:55:37 +08:00
2026-06-16 14:55:37 +08:00
2026-06-16 14:55:37 +08:00
2026-06-16 14:55:37 +08:00
2026-06-16 14:55:37 +08:00

DeepSeek Harness

Monorepo for the DeepSeek Harness group.

Projects

  • DeepSeek Code — DeepSeek's coding agent product.

Development

This monorepo is built on the Cordis framework (vendored as source under vendor/), microkernel-style: everything is a plugin.

pnpm install
pnpm run test          # vitest
pnpm run demo:echo     # runnable echo-agent example (no API key needed)
pnpm run demo:coding   # the real DeepSeek coding agent (needs DEEPSEEK_API_KEY)

For humans, start with the development guide for local setup, hooks, environment variables, and quality gates, then read the architecture design before package work. Local context lives in packages/ and vendor/.

For agents, follow AGENTS.md.

Description
No description provided
Readme MIT 120 MiB
Languages
TypeScript 97%
CSS 1.6%
Python 0.7%
JavaScript 0.6%