Wiki Ingest logo

Wiki Ingest

Community
Mark393295827
wiki-ingest

Use when a PDF, URL, transcript, clipping, or raw note must become source-grounded, linked, governed knowledge in an Obsidian vault.

Overview

PublisherMark393295827
Repositorythird-brain-v7-skills
Skill namewiki-ingest
Stars
141
Forks
22
Bundled files
1
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 1 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by Mark393295827 on GitHub. Read the source before you install it.

Installation

Install the Wiki Ingest AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/Mark393295827/third-brain-v7-skills/tree/main/skills/wiki-ingest
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/Mark393295827/third-brain-v7-skills.git /tmp/third-brain-v7-skills
mkdir -p .claude/skills
cp -r /tmp/third-brain-v7-skills/skills/wiki-ingest .claude/skills/wiki-ingest
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Wiki Ingest in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Wiki Ingest on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Wiki Ingest is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Wiki Ingest

<skill_contract> Accessible PDF, URL, transcript, clipping, or note plus vault root, topic, provenance, permissions, and lifecycle context. Immutable source evidence, linked concept and entity updates, navigation and lifecycle changes, and a verified ingest receipt. Touched files pass source, link, block-ref, frontmatter, lifecycle, understanding, and read-after-write checks. <non_goals>Fabricated provenance, source mutation, raw-summary dumping, unsupported rule promotion, or unverified completion.</non_goals>

Use STOW: Source -> Transform -> Organize -> Write-back. The source is evidence, the concept page is current understanding, and the log is the receipt. Load references/stow-contract.md for schemas and lifecycle details.

Usage Template

Provide: source path/URL/note, vault root, intended topic, permissions, and whether the item came from a clipping queue. Optional: path config, prior canonical source, and desired behavior/skill candidate.

Workflow

  1. Verify source existence, type, accessibility, date, and vault identity.
  2. Resolve configured directories; record defaults only when no config exists.
  3. Classify input as external fact, human experience, internal state, or environment signal.
  4. Assign source risk based on primary/secondary/mediated/self-reported status and freshness.
  5. Derive a source identity from canonical URL or source id, then hash/title/date fallbacks. Search canonical sources, concepts, and entities before creating files; record the identity query in the run ledger.

<unknowns_gate>

If the source cannot be read, the vault is ambiguous, or provenance cannot be distinguished from mediation, return NEEDS_INPUT or INSUFFICIENT_EVIDENCE. Never fabricate author, date, URL, hash, source id, transcript text, or primary-source confidence.

</unknowns_gate>

wiki-ingest owns one evidence-to-candidate transaction: immutable source capture, concept candidate, graph update plan, governance handoff, and clipping transition after verified commit. Cross-corpus deduplication, retrieval/index maintenance, and knowledge-debt queues belong to knowledge-ops; health scoring and generalized repair recommendations belong to wiki-lint.

  1. Source: create or identify one immutable source note; preserve raw content and archive locators. Extract 3-7 key insights with stable block references. For concurrent ingest, stage the candidate, recheck the source identity immediately before commit, and promote only one canonical path.
  2. Transform: separate direct claims, interpretations, contradictions, unknowns, and fast-changing facts. Mark single-source and self-reported claims.
  3. Organize: update existing entity/concept pages when possible. New concepts require a thesis, mechanism, boundary, counterpoint, source locator, and at least two meaningful links.
  4. Understand: answer what changed, what causes what, what this source may prove, what could make it wrong, and what reusable action follows. Raw summaries fail this gate.
  5. Navigate: update the relevant index/map and clipping lifecycle. Machine-owned snapshot blocks remain untouched.
  6. Convert: queue behavior, creativity, SOP, or skill candidates; apply the promotion gate before changing governed rules.
  7. Write-back: append evolution timeline and ingest log entries; never rewrite historical provenance. Give every machine log entry an idempotency key derived from run/batch, operation, canonical source identity, and target; a retry must not append the same logical receipt twice.
  8. Verify: run targeted link, source-ref, frontmatter, block-ref, empty-file, navigation, and post-write exact-identity checks over touched files. If a concurrent source appears, keep one canonical note, repoint derived links, and record reconciliation before claiming completion.

For material promotion or source ambiguity, require independent review and human approval. Prepare a rollback as a file-level diff; rollback may remove derived writes but must not erase original source evidence or logs.

Read touched files after write. Confirm the source exists, locators resolve, concepts pass the understanding gate, links are meaningful, clipping status is accurate, and logs match actual writes. Downgrade the result to partial when any required check is unavailable.

<retry_policy>

max_attempts: 2. Retry only the failed stage after diagnosing path, parser, schema, or link cause. Change strategy before retry; stop on the same signature or NO_PROGRESS. Never duplicate source notes as a retry mechanism.

</retry_policy>

<state_contract>

Persist {run_id, status, attempt, budget, evidence, unknowns, last_error, next_action} plus source identity/risk, idempotency key, writer lease or compare-and-set evidence, vault fingerprint, resolved paths, touched-file ledger, created block refs, contradictions, reconciliation decisions, candidates, approval, clipping transition owner, and verification receipts. Source and log history are append-only.

</state_contract>

Failure Protocol

  • NEEDS_INPUT: source or vault identity is ambiguous; ask one minimal probe.
  • INSUFFICIENT_EVIDENCE: the material cannot support the requested claim or concept.
  • BLOCKED_PERMISSION: return proposed writes and paths without claiming mutation.
  • VERIFY_FAILED: a post-ingest check fails; preserve evidence and repair only the failed stage.
  • NO_PROGRESS: repeated failure signature after changed strategy; stop and queue review.
  • BUDGET_STOP: persist coverage and remaining items. max_attempts: 2.

Output Contract

Return status, result (created/updated/queued files and understanding delta), evidence (source and lint receipts), unknowns, and next_action including approval or rollback when relevant.

Edge Cases

  • A clipping duplicates an existing source: link/archive it and add only new provenance or block locators; do not create a parallel canonical source.
  • Two workers discover no exact-URL match and race to create a source: the commit-time identity check selects one canonical note; remove only the uncommitted/staged duplicate, repoint derived links, and log reconciliation.
  • A retry sees its write-back idempotency key in the log: verify the existing receipt and return reused/no-op instead of appending another entry.
  • A mediated summary contains a precise changing number: preserve the summary, flag the number, and queue primary/current verification before promotion.

Success Metrics

  • Every derived claim traces to an immutable source locator.
  • Concurrent retries converge to one canonical source and one logical receipt.
  • Concepts add mechanism and boundary, not only summary.
  • Touched files pass targeted lint and appear in navigation or an explicit queue.

Quality Gates

  • Vault, source, and permissions were verified before write.
  • No provenance field or source content was invented.
  • Source identity was checked before and after write; log append was idempotent.
  • Understanding and promotion gates were applied.
  • Independent/approval/rollback controls match the change risk.
  • Post-ingest receipt matches actual filesystem state.

</skill_contract>

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Wiki Ingest AI skill do?

Use when a PDF, URL, transcript, clipping, or raw note must become source-grounded, linked, governed knowledge in an Obsidian vault.

Why use Wiki Ingest on TypingMind?

Because you install it once and use it with any model. Wiki Ingest is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Wiki Ingest in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/Mark393295827/third-brain-v7-skills/tree/main/skills/wiki-ingest. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Wiki Ingest?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Wiki Ingest?

As many as you like. As long as a model supports skills, you can use Wiki Ingest with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Wiki Ingest AI skill free?

Yes. It is published on GitHub by Mark393295827 under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇