Skip to content
Back to skills

Reconcile

ASecurity

Reconcile what the writer SAYS, what their page STATES, and what their corpus SHOWS — surface the gaps as open questions, capture the answers, and classify each as aspiration / drift / latent-strength / noise. Step 5 of the Context Kit — the engine that injects direction. Run after distill-subject, before the Brief. Use for "reconcile", "compare what I say vs what I write", "what's the gap between how I see it and what it is".

  • 2 stars
  • 0 votes
  • 0 copies
  • 1 view
  • Added September 27, 2026
ai-agents

Security analysis

A100/100

Scanned September 27, 2026

npx -y skills add Artifacts7/newsletter-context-kit --skill reconcile --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Reconcile?

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

Security grade badge for Reconcile
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/artifacts7-reconcile/badge)](https://www.skillsdirectory.com/skills/artifacts7-reconcile)

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: reconcile
description: >
  Reconcile what the writer SAYS, what their page STATES, and what their corpus SHOWS — surface
  the gaps as open questions, capture the answers, and classify each as aspiration / drift /
  latent-strength / noise. Step 5 of the Context Kit — the engine that injects direction. Run
  after distill-subject, before the Brief. Use for "reconcile", "compare what I say vs what I write",
  "what's the gap between how I see it and what it is".
---

# Reconcile (step 5) — the engine

Turn three static views into directional context. The corpus is descriptive (what the newsletter
*was*); this dialogue is the only place **aspiration** (what it *wants to be*) enters.

## Precondition
`newsletter-context/seed-context.md` (said), `newsletter-context/identity-stated.md` (stated), `newsletter-context/subject-observed.md` (shown).

## Procedure
1. **Three-way diff.** Lay the views side by side; find disagreements. Four gap types: said-not-shown · shown-not-said · stated-vs-shown · said-vs-stated.
2. **Prioritize** the top 3–5 gaps that most change the Brief (overwhelm kills the gate).
3. **Ask, don't assert** (Q7 — anchoring). One *open* question per gap — never leading. Present the agent's read as challengeable, not as the answer.
4. **Capture + classify** each answer: aspiration (→ objectives) · drift (→ down-weight/prune) · latent strength (→ surface) · noise (→ exclude).

## Note
This step is **interactive** — the writer's answers are the whole point. Do not resolve gaps for
them. If they skip one, mark it open.

## Output
`newsletter-context/reconcile.md` — gaps, questions, and resolutions. → hand to **`brief`** (step 6).

---
_**Language:** if the newsletter isn't in English, converse with the writer — and write outputs — in the newsletter's language._

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.