菜单
Use when stress-testing tensor-grep against a multi-project workspace — orientation, scoped search, tg find, prepare, route-test, ledger, install-dense, symbol graphs, GPU, evidence, sessions, readiness gates. Not the PyPI release dogfood harness.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/oimiragieo/tensor-grep --skill tensor-grep-workspace-dogfood命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
Use when searching code, logs, or repositories with tensor-grep; validating rg or AST parity; using tg MCP tools; checking GPU/search routing; or producing agent-friendly context, source, refs, or blast-radius output.
Use when writing or editing any tensor-grep doc of record (AGENTS.md, CLAUDE.md, root SKILL.md, docs/SESSION_HANDOFF.md, docs/CONTRACTS.md, docs/CONTINUATION_PLAN.md, docs/PAPER.md, README.md, docs/benchmarks.md, docs/gpu_crossover.md, mkdocs.yml, or any .claude/skills/*/SKILL.md) — before adding a capability claim, syncing a release-line note, touching a version-stamped line, adding a new governed doc, or a docs-only pytest fails and you don't know which fragment broke it. Covers which doc owns which contract, the two governance layers (semantic-release version_variables + scripts/stamp_release_assets.py auto-stamping vs pytest content-pinning), the cross-doc fragment discipline, mkdocs --strict, and house style.
Use when deciding what counts as proof that a tensor-grep (tg) change works — before trusting a subagent's "tests pass", writing a new test, claiming a routing/docs/release fix is done, shipping a doc-drift/ranking/classification heuristic off green fixture tests, proving a red-green baseline on a reverted/pre-fix commit, reviewing a payload-byte/ratio governance test, de-flaking a timing-sensitive test, or running the pre-push gate. Covers TDD-first discipline, the CliRunner-vs-real-binary trap, the fixture-green-vs-real-corpus-dogfood trap for precision/heuristic features, the `capfd`-vs-`result.stdout` capture-surface trap on routing/delegation changes (needs `tests/integration/` run with the native `tg` binary rebuilt, not just `tests/unit/`), the `tests/conftest.py` `sys.path.insert`-outranks-`PYTHONPATH` trap that can falsify a red-green baseline even with `tensor_grep.__file__` verified, a shared-envelope field growth breaking a payload-ratio governance test plus its tmp-path-length platform sensitivit
正在显示 SKILL.md
| name | tensor-grep-workspace-dogfood |
| description | Use when stress-testing tensor-grep against a multi-project workspace — orientation, scoped search, tg find, prepare, route-test, ledger, install-dense, symbol graphs, GPU, evidence, sessions, readiness gates. Not the PyPI release dogfood harness. |
# Prefer an explicit published pin when comparing skills to product:
uvx --from tensor-grep==1.110.14 tg --version
# Bare `uvx --from tensor-grep tg` / a shadowed `C:\Users\...\bin\tg` can report a stale version.
tg doctor --json ROOT
tg devices
cd /path/to/workspace
tg calibrate
tg search PATTERN tensor-grep/src --type py --gpu-device-ids 0 --json
tg agent tensor-grep/src "task" --gpu-device-ids 0 --gpu-timeout-s 15 --json | jq .gpu_acceleration
tg inventory tensor-grep --json
tg orient tensor-grep --ignore "node_modules/**" --json
tg search TODO . --glob "*.py" --max-depth 3 --json
tg find "session daemon timeout" tensor-grep/src --deadline 20 --json
# Prefer prepare over the multi-step agent loop for edit readiness:
tg prepare tensor-grep/src "task" --json
tg prepare tensor-grep/src "task" --claim --json
tg prepare tensor-grep/src "task" --out /tmp/capsule.json --json # persist for evidence emit --capsule
tg prepare tensor-grep "task" --deadline 20 --json # expect partial on whole-repo
tg route-test tensor-grep/src "task" --json
tg agent tensor-grep "task" --deadline 20 --json # still flaky without explicit deadline
tg evidence emit tensor-grep --capsule /tmp/capsule.json --query "task" --json --agent-id dogfood > /tmp/receipt.json
tg ledger claim|record|find|release … # see tensor-grep-ledger
tg install-dense --json # once; then re-try tg find
tg agent agent-studio/.claude/lib/routing "task" --json
tg dogfood --root . --output /tmp/dogfood-ws.json
| Category | Result | Notes |
|---|---|---|
| Core CUJ | ✅ | 21/21 PASS; prepare saddle ~5s; tg src --out/--claim ~22s |
| Parent refuse | ✅ | exit 2; workspace_root_refused |
| A90 reserved+flag refuse | ✅ | edit-ready/--json → exit 2 unknown_command; typo searhc --help suggests search |
| A90 bare reserved token | ✅ by design | tg edit-ready (no flag) still searches that string |
M16 scan / M17 --index | ✅ | retained from 1.110.12 |
| Language coverage | ✅ | 10/10 parser-backed |
Phase-2 edit-ready product | ❌ | correctly refused now; not implemented |
Artifact: C:\Users\Public\tg-dogfood-111013.json.
Version annotation (2026-08-12): this section was previously stamped "tg 1.110.14", but the
cited artifact does NOT cover 1.110.14 — it is decisively a 1.110.13 sweep: the artifact records
"version": "1.110.13" with "when": "2026-08-11T12:26:56Z" (BEFORE v1.110.14 was tagged,
2026-08-11 17:32 UTC), and its sweep script tg-dogfood-111013.js pins VER = '1.110.13'. No
1.110.14 sweep artifact exists in C:\Users\Public (only 111010/111012/111013). The header above
was corrected to the version the artifact actually covers; the table rows below are unchanged
because they match the artifact (counts.PASS = 21).
Kept for trend only. Parent refuse class was still recorded as scan_limit in that sweep —
superseded by workspace_root_refused on 1.110.x (#956).
| Version | PASS | TIMEOUT | Notable |
|---|---|---|---|
| 1.101.22 | saddle ✅ | — | bare---json PATH note (stderr) |
| 1.101.31 | saddle ✅ | — | bare---json in-band scope_note; MaxSim de-advertised |
| 1.108.2 | saddle ✅ | — | CUJ stable; parent refuse class then=scan_limit |
| 1.110.10 | saddle+tg ✅ | — | workspace_root_refused; 10/10 parser-backed; prepare~6–14s |
| 1.110.12 | saddle+tg ✅ | — | +M16 scan severity/message; +M17 index root isolation surface; CUJ 21/21 |
| 1.110.13 | saddle+tg ✅ | — | +A90 unknown-command fail-closed; CUJ 21/21 |
Table ends at 1.110.13 ON PURPOSE: the latest receipt artifact (tg-dogfood-111013.json) is a
1.110.13 sweep (see the version annotation above), and no 1.110.14 artifact exists. Do NOT add a
1.110.14 row until a genuine 1.110.14 sweep artifact exists — a trend row without an artifact is a
fabricated receipt.
tensor-grep, tensor-grep-prepare, tensor-grep-ledger, tensor-grep-find-and-route, tensor-grep-gpu, tensor-grep-enterprise-agent, tensor-grep-multi-project-search