amhuppert/claude/ai-resources-plugin/skills/kb-ingest
claude/ai-resources-plugin/skills/kb-ingest/SKILL.md
--- name: kb-ingest description: Delegate knowledge-base ingestion to the `knowledge-base-ingester` sub-agent so the source document, primary/current trees, and ingest workflow stay out of the main conversation's context window. Run this when the user explicitly invokes `/ai-resources:kb-ingest`; do not auto-trigger from natural-language ingestion requests (those should use the `knowledge-base-ingest` skill directly). disable-agent disable-model-invocation: true --- # kb-ingest Run a knowledge
$ install --global
skillsauthnpx skillsauth add amhuppert/my-ai-resources claude/ai-resources-plugin/skills/kb-ingestInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
Security Scan Results
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
3
Scanners Passed
9
Scanners in report
Clean
TrivyContainer and dependency vulnerability scanner
95%
Clean
SemgrepStatic code analysis for vulnerabilities
95%
Clean
mcp-scan (Snyk)Model Context Protocol security validation
95%
Skipped
Snyk (dep)Open source security scanning
50%
Skipped
Socket.devSupply chain security analysis
50%
Skipped
VirusTotalMulti-engine malware detection
50%
Skipped
CrowdStrikeAdvanced threat intelligence
50%
Skipped
OSV-ScannerOpen Source Vulnerability database check
50%
Skipped
OWASP Dep-Check
50%
Last scanned: May 17, 2026, 3:31 AM171.1s1 file scanned
SKILL.md
- name:
- kb-ingest
- description:
- Delegate knowledge-base ingestion to the `knowledge-base-ingester` sub-agent so the source document, primary/current trees, and ingest workflow stay out of the main conversation's context window. Run this when the user explicitly invokes `/ai-resources:kb-ingest`; do not auto-trigger from natural-language ingestion requests (those should use the `knowledge-base-ingest` skill directly).
- disable-model-invocation:
- true
kb-ingest
Run a knowledge-base ingestion in an isolated agent context.
This skill exists for one reason: ingesting a source document into a knowledge base loads a lot — the source itself, the primary/derived/current trees, the knowledge-base-ingest skill, and its references. When the parent agent runs it inline, that all lands in the main conversation's context window. This skill pushes the entire workflow into the knowledge-base-ingester sub-agent and brings back a compact report.
When to invoke
Only when the user runs /ai-resources:kb-ingest (with or without arguments). Do not auto-trigger on phrases like "ingest this document" or "add to the knowledge base" — those should run the knowledge-base-ingest skill directly in the main conversation. This skill is the explicit, user-chosen path for the delegated workflow.
What the skill does
- Parse the user's invocation for ingest inputs.
- Delegate the actual ingestion to the
knowledge-base-ingestersub-agent via the Agent tool. - Surface the agent's compact report to the user.
- Do not re-do the agent's work inline.
Step 1: Parse inputs
From the user's /ai-resources:kb-ingest invocation, gather:
- Source document path: the file to ingest. Required.
- Knowledge-base destination: explicit path (e.g.
kb/api-redesign/) if given; otherwise leave it to the agent's discovery rules. - Caller intent: pure ingestion, or ingestion + extra analysis the user asked for.
- Any extra instructions: scope limits, "don't touch X", deadlines, etc.
If the user supplied a path that does not exist, or supplied nothing at all, ask them for the source path before delegating. Do not invent a path.
If the user gave instructions but no source path, ask which document they want ingested. Do not guess.
Step 2: Delegate to the sub-agent
Invoke the knowledge-base-ingester sub-agent via the Agent tool with subagent_type: "knowledge-base-ingester". The prompt to the agent must be self-contained (the agent does not see the parent conversation) and must include:
- The exact source document path.
- The knowledge-base destination if the user specified one, or a note that destination should be inferred per the skill's discovery rules.
- Any user-supplied scope limits or extra analysis requests.
- A reminder to follow the
knowledge-base-ingestskill and return the report in the format defined by the agent.
Minimal prompt template:
Ingest the following source into a knowledge base, following the `knowledge-base-ingest` skill end-to-end. Return the compact report defined in your agent definition — do not paste source content.
Source: <absolute or repo-relative path>
Knowledge base: <explicit path, OR "infer per the skill's discovery rules">
Extra instructions: <anything the user said beyond "ingest this", or "none">
Run the agent in the foreground. The parent needs the report before it can respond to the user.
Step 3: Surface the report
When the agent returns, relay its report to the user with minimal added commentary. The agent's report already names file paths, the impact summary, and validation steps performed — do not paraphrase it into something longer.
Two things to do on top of the report:
- Flag escalations. If the agent returned an early-exit (ambiguous destination, unreadable source, etc.) instead of a full report, surface that clearly and ask the user the question the agent raised.
- Note unresolved limitations. If the agent's "Limitations / Assumptions" section is non-empty, call those out so the user can decide whether to follow up.
Step 4: Stop
Do not re-run the ingestion inline to "double-check." Do not re-read the primary documents the agent wrote. Do not summarize the source. The whole point of this skill is to keep that work out of the main context window — duplicating it in the parent defeats the purpose.
If the user follows up with edits, refinements, or a second source, invoke the sub-agent again rather than handling the work directly.
Anti-patterns
- ❌ Reading the source document into parent context "to understand it" before delegating. The agent reads the source; the parent should not.
- ❌ Reading the agent-written primary/current files after the agent finishes "to verify". The agent's report is the verification. If you don't trust it, the agent definition is the thing to fix, not this skill.
- ❌ Auto-triggering on natural-language ingest requests. Use the
knowledge-base-ingestskill directly in those cases. This skill is the explicit delegated path. - ❌ Repeating the agent's full report verbatim with your own paraphrase appended. Relay it; don't pad it.
Related Skills
amhuppert/hotkey-helper
tools
VerifiedTrustedCommunity
Use when picking or vetting a keyboard shortcut on macOS. Triggers include "what hotkey should I use for X", "is `<combo>` available", "does this shortcut conflict", "recommend a keybinding for…", "check `<combo>` against my setup", "pick a hotkey for…", or any mention of choosing/binding/changing a shortcut in WezTerm, tmux, Zed, Chrome, Claude Code, or macOS. Determines whether a proposed combo collides with OS-reserved bindings, app defaults, or the user's customizations, and recommends ergonomic alternatives when needed.
2SKILL.mdUpdated Jun 2, 2026
amhuppert/knip
development
VerifiedTrustedCommunity
Detect and remove dead code with knip. Use when the user asks to "run knip", "find unused files", "find unused exports", "find unused dependencies", "clean up dead code", "remove dead code", "set up knip", "configure knip", "knip.json", "knip false positive", "knip CI", or mentions a `knip` config, dependency bloat, bundle bloat from unused imports, or tree-shaking unused exports. Covers the configuration-first workflow, confidence-gated deletion, framework-specific gotchas (Next.js 15+, Tailwind, Storybook, Jest, Bun's test runner and `bun build --compile`), monorepos, CI integration, and performance tuning.
1SKILL.mdUpdated May 27, 2026
amhuppert/react-scan
tools
VerifiedTrustedCommunity
This skill should be used when the user asks to "set up react-scan", "install react-scan", "diagnose React re-renders", "find unnecessary renders", "find unstable props", "automate React render checks with Playwright", "react-scan + playwright", "measure component renders programmatically", "check why a React component is slow", or mentions React rendering issues, slow React interactions, render counts, or component-level perf attribution. Covers install across Next.js/Vite/Remix/script-tag/browser-extension, the lite headless API for CI, and the canonical render-attribution → fix → validate loop driven through Playwright.
1SKILL.mdUpdated May 20, 2026
amhuppert/knowledge-base-ingest
documentation
VerifiedTrustedCommunity
This skill should be used when integrating source material into a knowledge base, including when the user asks to "integrate this document into the knowledge base", "add this transcript to the memory bank", "ingest this document", "update the knowledge base", "analyze a new source document", or "sync current-state docs with this source".
1SKILL.mdUpdated May 17, 2026