AugmentClaude

Source Ingest

Analyze a URL or document snapshot into a structured knowledge base report.

Installation

  1. Make sure Claude is on your device and in your terminal.

    Skills load from ~/.claude/skills/ when Claude Code starts up — so you need it on your machine first. If you don't have it yet, install it once with the command below, then run claude in any terminal to verify.

    One-time setup
    npm i -g @anthropic-ai/claude-code

    Already have it? Skip ahead.

  2. Paste into Claude Code or into your terminal.

    This copies the whole skill folder into ~/.claude/skills/cp-skill-ingest-zby/ — the SKILL.md plus any scripts, reference docs, or templates the skill ships with. Safe default: works for every skill.

    Faster alternative (instruction-only skills)

    Skips the clone and grabs only the SKILL.md file. Don't use this if the skill ships Python scripts, reference markdowns, or asset templates — they won't be downloaded and the skill will fail when it tries to load them.

    Quick install (SKILL.md only)
    Sign up to copy
  3. Restart Claude Code.

    Quit and reopen Claude Code (or any other agent that loads from ~/.claude/skills/). New skills are picked up on startup.

  4. Just ask Claude.

    Skills auto-activate when your request matches the skill's description — no slash command needed. Trigger phrases live in the skill's own frontmatter; you can read them in the “What this skill does” section above.

Prefer to read the source first? Open on GitHub.

When Claude uses it

Use when asked to ingest a URL or existing kb/sources snapshot into a .ingest.md source analysis.

What this skill does

Ingest source

Ingest one source into a kb/sources/*.ingest.md report. The source may be a URL or an existing Markdown snapshot under kb/sources/.

Contract

Target: $ARGUMENTS

The direct output is the .ingest.md report next to the source snapshot. URL snapshotting and connection discovery may write their own delegated artifacts. Do not directly write any other library artifacts.

Interpret "our" through the installed KB's goals and local collection contracts. In this repository, "our" means agent-operated KB methodology. In another installed KB, it means that project's declared system, work, codebase, policy, product, or domain.

Read and follow kb/sources/types/ingest-report.md before drafting the report. If this skill and the type spec conflict about report content, the type spec wins.

Steps

  1. Resolve the target.

    • If $ARGUMENTS is empty, list recent kb/sources/*.md files excluding .json and .ingest.md, then ask which one to ingest.
    • If the target starts with http:// or https://, invoke cp-skill-snapshot-web on the URL. Parse the Snapshot saved: line from its output; that path is the source snapshot for the next step.
    • Otherwise, treat the target as the source snapshot path.
  2. Run connection discovery. Invoke cp-skill-connect on the source snapshot path. Wait for it to finish. For source snapshots, read kb/reports/connect/sources/<snapshot-name>.connect.md.

  3. Extract connection context. From the generated connect report, note:

    • Which existing artifacts were identified as connections
    • What relationship types were found
    • Any synthesis opportunities or tensions flagged

    Treat Maintenance Observations as non-actionable context: mention durable signals in the ingest report only when relevant, and do not act on or promote them during ingest.

    The connect report is generated, gitignored working context. Do not cite it, link to it, or name its path in the durable ingest report. Summarize its findings and link only durable KB artifacts or source snapshots.

    Distill, do not transcribe: connect casts a wide candidate net by design. Drop weak, speculative, or duplicate edges and keep only settled, durable judgments about this source's role. Write Connections Found as compact prose naming that role (for example: anchor, technical basis, counterpoint, legal disposition, public statement, limitation) rather than an inventory of every candidate connect surfaced.

    If no casebook notes exist yet for the target collection, say so plainly and stop there. Do not substitute a full map of this source's relationships to other already-captured sources, and do not frame the section as prospective connections for notes that do not exist yet — that framing goes stale the moment notes are written, and re-deriving those relationships is connect's job to do then, not the ingest's job to pre-write now. A single relationship that bears on this source's likely role is enough.

  4. Draft the ingest report. Write the analysis as an ingest-report, using the source snapshot and the connection context. The report must classify the source, summarize it, explain how it connects to the current KB, extract goal-relative value, state limitations, and recommend one advisory next action.

    If the source is not relevant to this KB, say so in the report. Keep the report short, explain the mismatch, and recommend no promotion or source-only filing as appropriate.

  5. Save the report next to the snapshot.

    • Input: kb/sources/some-article.md
    • Output: kb/sources/some-article.ingest.md
  6. Validate. Run:

    commonplace-validate kb/sources/some-article.ingest.md
    

    If this run created or edited the source snapshot, validate that snapshot too. Fix validation failures in files this skill is allowed to write before stopping.

  7. Report the result. Tell the user where the ingest report was saved and state the recommended next action.

Constraints

  • Run cp-skill-connect before classification or value extraction.
  • Write only the .ingest.md report directly.
  • Base extractable value on what is new relative to the discovered connection context.
  • Include effort tags on extractable value items.
  • Recommend exactly one advisory next action.

Related skills