Ingest Runner — Development agent for Claude Code
Executes the fast and lean ingest pipeline (skills/ingest/SKILL.md) for exactly one source — Python-first Raw capture, Source Note distillation, Entity creation, selective Concept extraction, and conc.
How to install Ingest Runner
Installs to ~/.claude/agents/migchw-llm-wiki-starter-ingest-runner.md
mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/Migchw/llm-wiki-starter/HEAD/.claude/agents/ingest-runner.md -o ~/.claude/agents/migchw-llm-wiki-starter-ingest-runner.md Restart Claude Code, or start a new session, for it to be picked up.
What Ingest Runner does
name: ingest-runner description: Executes the fast and lean ingest pipeline (skills/ingest/SKILL.md) for exactly one source — Python-first Raw capture, Source Note distillation, Entity creation, selective Concept extraction, and concise logging.
Ingest Runner — pipeline executor
Read `.agents/AGENTS.md` and `.claude/skills/ingest/SKILL.md` first — that skill is the single source of truth for the pipeline steps.
Execution Rules
- Python-First & Tiered Raw Capture: ใช้ `pytho
Alternatives in Development
- Wiki Ingest 3.3k ★
- Finding Linker — Compares new findings against a shortlist of existing findings in the vault and proposes typed edges (supports 95 ★
- Wiki Compile — Write-scoped wiki compile lane — spawn under an explicit lane contract (file whitelist, link whitelist, bounda 94 ★
Full documentation available on GitHub
View Source RepositoryRelated Agents
Ingest Concept Linker
Pass H cross-link pass of the 9-pass ingestion protocol. Reads the mechanical concept-candidates payload and d
Ingest Refs Extractor
Pass H refs pass of the 9-pass ingestion protocol. Reads one deep reference's header and thesis and writes its
Behavioral Story Extractor
Conducts STAR-extraction interviews to capture a single canonical story from the user's experience. Two modes:
Ticket Builder
Builds exactly one ticket of a planned feature by following the next-phase skill. Use it to run /next-phase in
Chief Editor
Receives every proposed concept and the story-doctor's report on each, picks the best ones looking for variety
Submission Runner
In an attended session, submit only exact application packages freshly confirmed by the user, then capture con
Related Skills
Memory Ingest
Process new raw content into the agent memory knowledge base. Creates source/concept/entity pages, cross-links
Synapse Ingest
Ingest a new monthly batch of academic papers into Synapse — runs the full pipeline (populate_manifest → lint_
Wiki Timeline
Generate a chronological timeline for an entity or concept in the LLM Wiki.