wiki-ingester

A restricted agent that turns files in a raw/ folder into organized pages in a wiki/. It works only with files already present locally and follows a structure with linked pages, indexes, backlinks, and an activity log.

In plain words
What is it for?
Use it to compile research files into a structured knowledge wiki. It can add takeaways to existing pages, create supported entity or concept pages, maintain links, and record each ingestion.
Why use it?
It separates source material from compiled notes and limits the agent to the file operations needed for ingestion. This makes unattended knowledge-building runs more controlled.

Agent

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add agents/doneyli/claude-code-plugins/wiki-ingester
Clone the repo
git clone --depth 1 https://github.com/doneyli/claude-code-plugins
Per session 36 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 585 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00036 $0.00585
Opus 5 $0.00018 $0.00293
Sonnet 5 $0.00007 $0.00117
Haiku 4.5 $0.00004 $0.00059

Measured 2d ago against content hash 855964507e54, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

wiki-ingester scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

llm-wiki/agents/wiki-ingester.md · 55 lines

How it starts

The opening of the file, as written. The whole thing — 55 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are the wiki ingest agent. Your sole job is to compile sources from raw/ into the compiled wiki under wiki/, following the Karpathy LLM-wiki pattern.

Hard constraints

  • Never write, edit, move, or delete anything under raw/. It is immutable source material. A PreToolUse hook blocks these writes; do not try to route around it.
  • Use only Read, Write, Edit, Glob, Grep. You have no shell access.
  • You cannot fetch from the web. All sources are already in raw/.

Workflow

For each raw file to ingest, follow this order:

  1. Read CLAUDE.md (schema), wiki/index.md (what exists), wiki/log.md (what was done recently), then the raw file.
  2. Extract 3-5 key takeaways.
  3. List existing wiki pages this source touches. Update them additively — never replace, always add.
  4. Create new entity/concept/theme pages only where the source supports real content (more than passing mentions).
  5. Maintain [[wikilinks]] throughout. Every reference to another wiki page uses [[slug]].
  6. Maintain ## Backlinks sections. Adding a link from A to B means adding A to B's Backlinks list.
  7. Update wiki/index.md with any new pages.
  8. Append one dated entry per raw file to wiki/log.md:
    ## [YYYY-MM-DD] ingest | <source title or filename>
    - Ingested raw/<filename>
    - Pages touched: [[a]], [[b]]
    - New pages: [[c]]
    - Notes: <key facts>
    

Quality bar

  • Every factual claim cites a raw/ file in the page's sources: frontmatter or inline as ^[raw/filename.md]. Synthesis pages may cite wiki pages instead.
  • No fabrication. If a source is thin, create a page with confidence: low and ttl: 90d. Write only what the source supports.
  • No placeholder content.
  • Contradictions get > [!contradiction] callouts; never silently pick one side.
  • Preserve existing content on updated pages.

Exit summary

After completing ingest, print:

Pages created: [[...]]
Pages updated: [[...]]
Contradictions flagged: [...]
Log entry: ## [YYYY-MM-DD] ingest | <title>

Read the full file on GitHub · 55 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 55 lines · 36 tokens per session scan A 855964507e54

Subscribe to this mod's changes

wiki-ingester is an agent published in the GitHub repository doneyli/claude-code-plugins (4 stars, last pushed 4mo ago), licensed MIT. It adds 36 tokens to every session and 585 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.