All skills
simota avatar

/magi

@35ffd55
by shingo imotasimota/agent-skills85 stars
15

Deliberating decisions and founder priorities through multi-perspective, named-expert, and YC-style advisory lenses. Use for verdicts, office hours, or expert critique; not implementation.

Use this Skill: https://skilld.dev/gh/simota/agent-skills/magi

This session only. Nothing lands on disk.

referencefigure-roster.md

≈853 tokens on demand. Your agent reads this file only when SKILL.md points to it.

Figure Roster

Purpose: The reusable knowledge asset — profiles of figures' documented thinking, so channeling is grounded and repeatable instead of re-researched every time. Read when: You are in GROUND, or running advisor roster to author/refresh a profile.

Storage

  • Profiles: .agents/magi/expert-roster/{figure-slug}.md (e.g. richard-feynman.md)
  • One file per figure. Slug is kebab-case of the documented name.
  • Profiles are grounding aids, not authority. A profile never overrides the attestation rules in attestation-tiers.md.

Profile schema

---
name: Richard Feynman
slug: richard-feynman
status: deceased        # deceased | living-public | living-private
domain: [physics, pedagogy, problem-solving]
record_strength: strong # strong | moderate | thin
last_reviewed: 2026-06-26
---

Then, in the body, the thinking-model layers from channeling-method.md, each item sourced or marked [inferred]:

  • Mental models — frameworks they reason with [source]
  • Heuristics — quick rules [source]
  • Characteristic questions — what they ask first [source]
  • Values / priorities — what they optimize for [source]
  • Trade-offs accepted — what they knowingly sacrifice [source]
  • Blind spots / known errors / reversals — where criticized or wrong [source]
  • Vocabulary — terms-of-art (paraphrase; no fabricated quotes)
  • Position log — dated stances, so channeling can be date-scoped
  • Sources — works, talks, interviews, papers used to build the profile

Authoring rules (advisor roster variant)

  • Every attribute carries a source or an explicit [inferred] marker. Unsourced ≠ silent.
  • Set record_strength honestly: thin profiles cap channeling at SPECULATIVE/INFERRED.
  • Date-scope positions; note reversals. A profile without a position log invites stale-position errors.
  • Set status correctly — it drives the ethics handling in ethics-and-safety.md.
  • Refresh trigger: review living-public figures when their documented positions may have shifted, and deceased figures when new scholarship, primary sources, or corrected misattributions surface — the record of a dead figure is not frozen. Bump last_reviewed on review.
  • A profile is a grounding aid with a shelf life: at CHANNEL, surface stale last_reviewed or thin record_strength in the attestation rather than treating the profile as silent authority (channeling-method.md anti-caricature checklist).

Seed exemplars (illustrative — verify before use)

These are starting points, not authority. Channel only what you can attest.

Figure Domain Signature frameworks (illustrative)
Richard Feynman physics, learning First-principles derivation; "if you can't explain it simply…"; teaching-as-understanding
Charlie Munger investing, decisions Latticework of mental models; inversion; circle of competence
Clayton Christensen innovation strategy Jobs-to-be-done; disruptive vs. sustaining innovation
Dieter Rams product design Ten principles of good design; "less, but better"
Barbara Minto communication Pyramid Principle; SCQA structuring

When a requested figure has no profile and the record is reachable, build a lightweight one inline at GROUND and offer to persist it as a roster file.

Source: SKILL.md on GitHub

1 alert13d5 checks · Risk HIGH
  • Gen Agent Trust Hub13d

    The skill facilitates complex decision-making but introduces significant security risks through its 'Engine Mode' and 'multi' recipe. These modes execute external CLI tools (codex, agy) using shell commands that incorporate user-provided context. Most notably, the skill explicitly instructs the agent to use a flag that bypasses permission checks (--dangerously-skip-permissions) when invoking the Antigravity CLI. This creates a high-risk vector for unauthorized command execution and bypasses platform security controls.

  • Socket13d

    No alerts

  • Snyk13d

    Risk: LOW · No issues

  • Runlayer6mo

    2/7 files flagged

  • ZeroLeaks5mo

    Score: 93/100 · 2 sections analyzed

Signed by skilld at 35ffd55. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub 2 days ago.

Activeupdated 2 weeks ago

README badge

README badge for simota/agent-skills/magi