AGENT SKILL

evolve-step

Review an existing Seed for blind spots, ask two blind-spot questions, and apply only user-approved changes to create the next Seed. The latest Seed replaces the top-level ## Seed section in .symposium/scratch/socrates.md; audit history is appended to .symposium/scratch/evolve-step.md. Triggers (EN): evolve-step, evolve seed, seed blind spots, seed v1 v2, review blind spots Do NOT use when: creating the first Seed from a vague request (-> socrates), running convergence through ontology (-> interview-harness), or only defining semantic boundary (-> ontology) vs socrates: Socrates creates Seed v1; Evolve-step turns an existing Seed into a user-approved next Seed.

License 许可
MIT
所属仓库
agisota/old-one
最近更新
2026-07-04
Stars
★ 3
Forks
1

01 前言 Front Matter

name
evolve-step
description
Review an existing Seed for blind spots, ask two blind-spot questions, and apply only user-approved changes to create the next Seed. The latest Seed replaces the top-level ## Seed section in .symposium/scratch/socrates.md; audit history is appended to .symposium/scratch/evolve-step.md. Triggers (EN): evolve-step, evolve seed, seed blind spots, seed v1 v2, review blind spots Do NOT use when: creating the first Seed from a vague request (-> socrates), running convergence through ontology (-> interview-harness), or only defining semantic boundary (-> ontology) vs socrates: Socrates creates Seed v1; Evolve-step turns an existing Seed into a user-approved next Seed.

02 资源文件 2

03 SKILL.md 正文

Evolve-step

Details

Evolve-step asks the model to propose blind spots, then lets the user decide what enters the Seed. The model never auto-adopts its own suggestions.

Input

Read the latest Seed from .symposium/scratch/socrates.md top-level ## Seed, or accept an explicit Seed YAML:

goal: ...
constraints:
  - ...
acceptance_criteria:
  - ...

Process

  1. Preserve v1 exactly.
  2. Generate exactly six candidates:
    • Q1: three user perspectives missing from the Seed.
    • Q2: three possibilities cut off by the Seed.
  3. Ask the user to mark each candidate as accepted or rejected.
  4. Apply only accepted candidates to Seed v2.
  5. Replace .symposium/scratch/socrates.md top-level ## Seed with Seed v2.
  6. Append the audit block to .symposium/scratch/evolve-step.md.

User Adoption Gate

Ask:

Mark each of the six candidates as accepted or rejected. No candidate is accepted by default.

Prefer a compact checkbox-style multiple-choice prompt:

Which candidates should enter Seed v2?
A. [ ] <missing user perspective 1>
B. [ ] <missing user perspective 2>
C. [ ] <missing user perspective 3>
D. [ ] <cut-off possibility 1>
E. [ ] <cut-off possibility 2>
F. [ ] <cut-off possibility 3>
G. None
H. Other / describe a different change

The user may answer with letters, checked boxes, or free-form edits. Accept only explicitly selected items.

Do not derive v2 before the user responds.

If no candidates are accepted, v2 equals v1 and the transferred meaning is No transferred meaning.

Audit Format

# Evolve-step — Cycle N

## Seed v1
```yaml
...

Blind Spot Candidates

Q1. Missing user perspectives

  • accepted/rejected — ...
  • accepted/rejected — ...
  • accepted/rejected — ...

Q2. Cut-off possibilities

  • accepted/rejected — ...
  • accepted/rejected — ...
  • accepted/rejected — ...

Seed v2

...

Transferred Meaning

...


## Rules

- Preserve v1 exactly in the audit.
- Apply only accepted candidates.
- Do not change multiple Seed slots for one candidate unless the user explicitly asks.
- Keep the transferred meaning concrete: name which slot changed and how.

04 同仓库更多 Skills

agisota
agisota/old-one

zoom-out

Tell the agent to zoom out and give broader context or a higher-level perspective. Use when you're unfamiliar with a section of code or need to understand how it fits into the bigger picture.

★ 3
agisota
agisota/old-one

zeroize-audit

Detects missing zeroization of sensitive data in source code and identifies zeroization removed by compiler optimizations, with assembly-level analysis, and control-flow verification. Use for auditing C/C++/Rust code handling secrets, keys, passwords, or other sensitive data.

★ 3
agisota
agisota/old-one

yeet

Use only when the user explicitly asks to stage, commit, push, and open a GitHub pull request in one flow using the GitHub CLI (`gh`).

★ 3
agisota
agisota/old-one

yara-rule-authoring

Guides authoring of high-quality YARA-X detection rules for malware identification. Use when writing, reviewing, or optimizing YARA rules. Covers naming conventions, string selection, performance optimization, migration from legacy YARA, and false positive reduction. Triggers on: YARA, YARA-X, malware detection, threat hunting, IOC, signature, crx module, dex module.

★ 3
agisota
agisota/old-one

x-twitter-growth

X/Twitter growth engine for building audience, crafting viral content, and analyzing engagement. Use when the user wants to grow on X/Twitter, write tweets or threads, analyze their X profile, research competitors on X, plan a posting strategy, or optimize engagement. Complements social-content (generic multi-platform) with X-specific depth: algorithm mechanics, thread engineering, reply strategy, profile optimization, and competitive intelligence via web search.

★ 3
agisota
agisota/old-one

x-api

X/Twitter API integration for posting tweets, threads, reading timelines, search, and analytics. Covers OAuth auth patterns, rate limits, and platform-native content posting. Use when the user wants to interact with X programmatically.

★ 3