📦 deps(thirdparty): update snapshots

2026-08-24 09:30:03 +08:00
parent 14e7209e78
commit 5d88510274
153 changed files with 3428 additions and 1203 deletions
@@ -0,0 +1,60 @@
+# DeepSeek Harness (dsh) Setup
+
+[DeepSeek Harness](https://github.com/deepseek-ai/deepseek-harness) (`dsh`) is DeepSeek AI's
+open-source agent harness — an "everything is a plugin" architecture with a Web UI, started with
+`npx @deepseek-ai/dsh web`. It natively loads [Agent Skills](https://agentskills.io) and reads
+`AGENTS.md`, so all six brooks-lint modes run with no conversion.
+
+## Install
+
+```bash
+# simplest — one command (global)
+curl -fsSL https://raw.githubusercontent.com/hyhmrright/brooks-lint/main/scripts/install.sh | bash -s -- dsh
+
+# from a clone
+./scripts/install.sh dsh            # global: ~/.dsh/skills (or $DSH_HOME/skills)
+./scripts/install.sh dsh --project  # this repo: ./.dsh/skills
+```
+
+Prefer a manual copy? Clone the repo and `cp -r skills/* ~/.dsh/skills/` — the contents, not the
+`skills/` folder itself, so `_shared/` lands as a sibling of the `brooks-*` folders.
+
+dsh scans these skill roots, highest priority first, so an existing vendor-neutral install is picked
+up automatically:
+
+| Root | Notes |
+|---|---|
+| `<projectRoot>/.dsh/skills` | what `--project` writes |
+| `<projectRoot>/.agents/skills` | shared with Cursor, Copilot, pi |
+| `$DSH_HOME/skills` (default `~/.dsh/skills`) | what the global install writes |
+| `$DSH_AGENTS_HOME/skills` (default `~/.agents/skills`) | `./scripts/install.sh agents` also covers dsh |
+
+The project root is the nearest ancestor containing `.git`; without one, dsh uses the current
+directory. When the same skill name appears in two roots, the higher one wins.
+
+## Invoke
+
+Just ask — dsh routes to a skill from its `description`:
+
+- "review this PR" → `brooks-review`
+- "audit the architecture" → `brooks-audit`
+- "where's our worst tech debt?" → `brooks-debt`
+
+For explicit invocation, type `/` in the Web UI prompt and pick the skill, or type the token by hand:
+`/brooks-review`, `/brooks-audit`, `/brooks-debt`, `/brooks-test`, `/brooks-health`, `/brooks-sweep`.
+dsh recognises a whitespace-bounded `/name` anywhere in a message and injects that skill's body
+deterministically. The repo's `AGENTS.md` carries the Iron Law (Symptom → Source → Consequence →
+Remedy) and the Health Score rules; dsh also loads `$DSH_HOME/AGENTS.md` plus every `AGENTS.md` from
+the project root down to your working directory.
+
+## Notes
+
+- **Flat layout** is mandatory (the installer guarantees it): discovery is one level deep
+  (`<root>/<name>/SKILL.md`), never recursive, and the skills read `../_shared/`, which only resolves
+  when `_shared/` sits beside the `brooks-*` folders. `_shared/` itself has no `SKILL.md`, so dsh
+  ignores it as a skill and the modes read it as ordinary files.
+- dsh is in developer preview and warns of compatibility-breaking changes; the skill discovery
+  contract above reflects `packages/skill/skill-filesystem` as of August 2026.
+- 🧪 Documented per the official [skills subsystem](https://github.com/deepseek-ai/deepseek-harness/blob/master/docs/subsystems/skills.md)
+  docs; community end-to-end verification welcome —
+  [open an issue](https://github.com/hyhmrright/brooks-lint/issues/new).
@@ -33,10 +33,10 @@ layout wrong:
 curl -fsSL https://raw.githubusercontent.com/hyhmrright/brooks-lint/main/scripts/install.sh | bash -s -- <platform>
 ```

-`<platform>` ∈ `opencode · cursor · windsurf · antigravity · pi · kiro · copilot · droid · gemini ·
-codex · claude · agents`. Add `--project` to install into the current repo instead of your global
-config. `agents` targets the vendor-neutral `~/.agents/skills` folder that Cursor, Copilot, pi, Gemini,
-and Codex all read.
+`<platform>` ∈ `opencode · cursor · windsurf · antigravity · pi · kiro · copilot · droid · dsh ·
+gemini · codex · claude · agents`. Add `--project` to install into the current repo instead of your
+global config. `agents` targets the vendor-neutral `~/.agents/skills` folder that Cursor, Copilot, pi,
+Gemini, Codex, and DeepSeek Harness all read.

 ## Per-platform guides

@@ -50,6 +50,7 @@ and Codex all read.
 | GitHub Copilot | [copilot-setup.md](copilot-setup.md) | `.github/skills`, `.claude/skills`, `~/.copilot/skills` | ✅ |
 | Kiro | [kiro-setup.md](kiro-setup.md) | `.kiro/skills`, `~/.kiro/skills` | ✅ |
 | Factory Droid | [factory-droid-setup.md](factory-droid-setup.md) | `~/.factory/skills`, `.factory/skills`, `.agent/skills` | ✅ |
+| DeepSeek Harness | [dsh-setup.md](dsh-setup.md) | `.dsh/skills`, `.agents/skills`, `~/.dsh/skills`, `~/.agents/skills` | ✅ |

 For Claude Code, Gemini CLI, and Codex CLI, see the [README install section](../README.md#installation).
 `./scripts/install.sh gemini` and `./scripts/install.sh codex` also work and use the flat layout these
@@ -68,7 +69,7 @@ If your agent accepts a skills folder or an instruction file, brooks-lint works:
 ## Verification status

 The marketplace-installed platforms (Claude Code, Gemini CLI, Codex CLI) are maintainer-verified. The
-eight Agent-Skills platforms above are documented from each tool's official skill spec and verified at
+nine Agent-Skills platforms above are documented from each tool's official skill spec and verified at
 the file-layout level (the installer is tested), but not yet end-to-end run by the maintainer on every
 platform. **Tried one? Tell us** — [open an issue](https://github.com/hyhmrright/brooks-lint/issues/new)
 with the platform, version, and what you saw, working or broken.
@@ -28,3 +28,4 @@ jobs:
          # fail-on: critical            # fail on any Critical finding (none | warning | critical)
          # fail-on-regression: true     # fail if the Health Score dropped vs the last run
          # sarif-file: brooks-lint.sarif  # also upload findings to GitHub Code Scanning
+          # api-base-url: https://your-gateway.example.com  # any Anthropic-compatible /v1/messages endpoint
@@ -28,7 +28,7 @@
  "description": "AI code reviews grounded in twelve classic software engineering books. Decay-risk diagnostics with book citations, severity labels, and six analysis modes including full-sweep auto-fix.",
  "url": "https://hyhmrright.github.io/brooks-lint/",
  "image": "https://hyhmrright.github.io/brooks-lint/hero.png",
-  "softwareVersion": "1.4.3",
+  "softwareVersion": "1.5.0",
  "license": "https://github.com/hyhmrright/brooks-lint/blob/main/LICENSE",
  "codeRepository": "https://github.com/hyhmrright/brooks-lint",
  "keywords": "AI code review, code quality, tech debt, architecture audit, test quality, Claude Code plugin, refactoring, clean architecture",