Skip to content
Back to skills

Manifest Core

ASecurity

Use when running any manifest-* advisor skill — the shared gather, prioritize, present, and route spine they all follow. Invoked by those skills, not directly by the user.

  • 2 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added September 4, 2026
ai-agents

Works with

  • cli

Security analysis

A100/100

Scanned September 4, 2026

npx -y skills add cavi-ai/obsidian-agent --skill manifest-core --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Manifest Core?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Manifest Core
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/cavi-ai-manifest-core/badge)](https://www.skillsdirectory.com/skills/cavi-ai-manifest-core)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: manifest-core
description: Use when running any manifest-* advisor skill — the shared gather, prioritize, present, and route spine they all follow. Invoked by those skills, not directly by the user.
---

# Manifest core

The spine every `manifest-*` advisor follows. The calling skill supplies only its lens
and its operationalizer.

**REQUIRED SUB-SKILL:** obsidian-agent:vault-grounding

## Process

1. **Gather.** Invoke `obsidian-agent:vault-synthesis` over the notes the lens
   names. Do not substitute uncited recollection; the synthesis keeps claims
   traceable to notes read through the official CLI.
2. **Apply the lens.** The calling skill defines what to look for and how to rank it.
   Tie every item to the notes that motivate it.
3. **Be opinionated.** Rank the candidates and lead with the single best next move. A flat
   unranked list is a failed output.
4. **Present.** Return portable Markdown: the top pick first, then the ranked
   list with rationale and evidence, followed by risks or unknowns. Use a table
   or fenced Mermaid diagram only when it improves comprehension.
5. **Operationalize.** For the items the user picks, invoke the operationalizer the
   calling skill names. Stopping at advice is a failed output.

## Hard requirements

- Every item traces to cited notes; mark inferences as inference.
- The output is a ranked Markdown brief leading with one clear next move.
- The run ends by routing into the named operationalizer, not at the brief.

## Common mistakes

- Generic domain advice ungrounded in the vault.
- No prioritization, or no single clear next move.
- Reading notes ad hoc instead of invoking vault-synthesis.
- Stopping at ideas instead of routing onward.

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.