본문으로 건너뛰기
cccskills
무료GitHub에 공개

dsh-ingest

Rebuild the DSH node's persistent knowledge state. Runs register-claude.py to (re)create the dome.db schema and upsert agents/skills/tools/stack rows, then runs ingest.py to re-index kb/, logs/, docs, and agent core source into ChromaDB's dome-kb collection, then runs pre-spore-verify.py to confirm all 7 skill modules load and the akashic write+query roundtrip works. Idempotent — safe to run any time.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

설치 방법 보기

포함된 파일(1)

  • SKILL.md5.1 KB

SKILL.md(원문)

설치하기 전에 에이전트에게 전달되는 지시 내용을 확인할 수 있습니다.

You are restoring or refreshing the DSH node's knowledge graph and agent registry. This skill is the recovery path when db/dome.db gets corrupted, when new KB content is added, or after an ingest-breaking dependency upgrade.

1. Preflight

cd "$DOME_ROOT"
test -f .venv/bin/activate || { echo "no venv — run dsh-setup first"; exit 1; }
source .venv/bin/activate
python3 -c "import sqlite3, chromadb, sentence_transformers, hashlib" || { echo "deps missing — pip install -r compute/requirements.txt"; exit 1; }

If the venv doesn't exist or deps are missing, stop and point the user at the dsh-setup skill first.

2. Register Claude + skills + tools in dome.db

python3 scripts/register-claude.py

What it does:

  • Creates (if missing) tables: sessions, stack, agents, skills, tools
  • UPSERTs the Claude agent row into agents
  • UPSERTs 8 skill rows into skills (docx, pdf, pptx, xlsx, schedule, setup-cowork, skill-creator, consolidate-memory)
  • UPSERTs 26 tool rows into tools
  • Mirrors everything into stack as category-tagged entries

Keyed on (category, name) so it's safe to re-run; only the updated_at and descriptions change.

Verify:

sqlite3 db/dome.db "SELECT name FROM sqlite_master WHERE type='table'"    # 5+ tables
sqlite3 db/dome.db "SELECT COUNT(*) FROM agents"                          # 1
sqlite3 db/dome.db "SELECT COUNT(*) FROM skills"                          # ≥ 8
sqlite3 db/dome.db "SELECT COUNT(*) FROM tools"                           # ≥ 26
sqlite3 db/dome.db "SELECT COUNT(*) FROM stack"                           # ≥ 35

If the skill count is higher than 8, new skills were registered elsewhere — that's fine, the upsert doesn't delete anything.

3. Re-index the knowledge base

python3 scripts/ingest.py

What it does:

  • Walks kb/, logs/, README.md, INDEX.md, MANUAL.md, agents/core/
  • Accepted extensions: .md, .txt, .py, .ts, .json
  • Skips .venv/ and __pycache__/
  • Chunks each file and embeds via sentence-transformers (all-MiniLM-L6-v2)
  • Writes to ChromaDB namespace dome-kb at db/chroma/

The script prints a per-file chunk count and a final total. Typical fresh-install expectation: 1,800–3,700 chunks depending on how many session logs are present.

Verify:

python3 -c "
import chromadb
c = chromadb.PersistentClient('db/chroma')
print(f'dome-kb: {c.get_collection(\"dome-kb\").count()} chunks')
"

4. Akashic collection hygiene

The akashic namespace uses a specific sentence-transformer embedding function. If it was previously created with a different embedder, the akashic:write+query roundtrip in pre-spore-verify will fail with Embedding function conflict.

Check:

python3 -c "
import chromadb
c = chromadb.PersistentClient('db/chroma')
names = [col.name for col in c.list_collections()]
print('collections:', names)
"

If akashic is in the list but empty (0 chunks) AND pre-spore-verify fails on it, drop and recreate:

python3 -c "
import chromadb
c = chromadb.PersistentClient('db/chroma')
if 'akashic' in [col.name for col in c.list_collections()]:
    c.delete_collection('akashic')
    print('dropped stale akashic collection — will recreate on first write')
"

The next pre-spore-verify.py run will seed the first akashic record (test write) cleanly.

5. Verify everything loads

python3 scripts/pre-spore-verify.py

Expected output ends with:

═══════════════════════════════════════
  27/27 checks passed

  ✅ ALL SYSTEMS GO — READY FOR SPORE.SH
═══════════════════════════════════════

The 27 checks:

  • 12 dependency imports (numpy, scipy, sympy, numba, torch, qiskit, pennylane, cirq, matplotlib, networkx, mpmath, pandas)
  • 7 skill-module verify() roundtrips
  • 1 akashic write+query roundtrip

6. Report

✅ dome.db: <agents>/<skills>/<tools> rows
✅ chroma/dome-kb: <N> chunks
✅ chroma/akashic: <M> chunks
✅ pre-spore-verify: 27/27

Non-negotiables

  • Never run ingest.py from outside $DOME_ROOT. It uses repo-relative paths; wrong cwd = wrong source set.
  • Never drop dome.db. The schema migrations are forward-only; there's no rollback. If the file is suspected corrupt, back it up (cp db/dome.db db/dome.db.bak.$(date +%s)) before deleting.
  • Never drop dome-kb blindly. Re-ingest takes 30+ seconds and you lose any manual records. Only delete when a schema or embedder change forces it.
  • Never commit db/chroma/ or db/*.db. Both are in .gitignore; they are per-node state, not code.

리뷰

아직 리뷰가 없습니다. 사용해 본 소감을 남겨 주세요.

댓글

로그인 없이 쓸 수 있습니다. 사용 후기, 질문, 작동하지 않은 환경 등을 편하게 남겨 주세요.

아직 댓글이 없습니다. 첫 댓글을 남겨 보세요.

보내면 바로 공개됩니다. 이메일 주소나 전화번호 같은 개인 정보는 쓰지 마세요. URL은 2개까지입니다.

0 / 1,000자

이 양식은 reCAPTCHA로 보호되며 Google 개인정보처리방침 및 서비스 약관이 적용됩니다.

같은 저장소의 스킬

ai-sdk

무료

Vercel AI SDK expert guidance. Use when building AI-powered features — chat interfaces, text generation, structured output, tool calling, agents, MCP integration, streaming, embeddings, reranking, image generation, or working with any LLM provider.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

garochee33/DSH42026년 10월 19일 업데이트

Algorithm design and analysis skill for DSH; use when working on optimization, data structures, or computational procedures.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

garochee33/DSH42026년 10월 19일 업데이트

Design and implement scalable API gateways, RESTful APIs, GraphQL endpoints, WebSocket handlers, and microservice communication patterns. Use when creating API routes, designing endpoints, implementing middleware, or setting up service mesh architectures.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

garochee33/DSH42026년 10월 19일 업데이트

best-of-n

무료

Implement a task N ways in parallel and pick the best. Spawns multiple subagents in isolated worktrees, evaluates all candidates, and applies the winner. Use when asked to "best of n", "try multiple approaches", "parallel implementations", "/best-of-n", or "/bon".

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

garochee33/DSH42026년 10월 19일 업데이트

check

무료

Check your work with a verification subagent. Spawns a verifier that reviews diffs, runs builds and tests, and evaluates correctness. Use when asked to "check work", "verify changes", "self-verify", "/check", "/verify", "/check-work", or "/self-verify".

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

garochee33/DSH42026년 10월 19일 업데이트

CI/CD pipeline design, GitHub Actions workflows, deployment automation, infrastructure as code, and release management. Use when setting up CI/CD, automating deployments, configuring build pipelines, or managing releases.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

garochee33/DSH42026년 10월 19일 업데이트

garochee33의 스킬 모두 보기

이 스킬의 문제 신고