himanshuj16/frank/frank-help

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Quick reference card for Frank: the commands, the modes, the three pushback shapes and the receipt format. One-shot display, changes nothing. Trigger: /frank-help, "frank help", "what does frank do", "how do I use frank".

Install

npx vskill@latest install himanshuj16/frank/frank-help

SKILL.md

Frank help

Print this card when invoked. One-shot: do not change the mode or write anything to disk.

Commands

CommandWhat it does
`/frank [lite\full\ultra\off]`Set the intensity, or report it with no argument.
/frank-verifyFind and run the verification for the last change, then write the receipt.
/frank-review [target]Check a transcript or PR description for openers, unverified claims, invented specifics and caves.
/frank-statsWhat Frank has caught, this session and all time.
/frank-helpThis card.

Codex invokes skills with @ (@frank ultra, @frank-review). Copilot CLI namespaces them (/frank:frank ultra).

Modes

ModeWhat changes
offNothing injected. Hooks stay quiet.
liteRules injected. The receipts gate reports but never interrupts.
fullDefault. Rules injected every turn and into subagents. The gate asks once per turn for a receipt.
ultraThe gate asks twice, and blocks a message that opens with flattery.

/frank with no argument shows a picker and says where the current mode came from. A switch sticks until the next one; /frank default hands control back to the configured default. That default is the plugin's Mode setting in Claude Code (asked for when Frank is enabled; change it with claude plugin install frank@frank --config mode=ultra), or on other hosts FRANK_DEFAULT_MODE or "mode" in ~/.config/frank/config.json (%APPDATA%\frank\config.json on Windows). FRANK_MODE in the environment overrides everything for that process.

The three shapes on pushback

  • HOLD: "Still <verdict>. Because <evidence>. What would change my mind: <thing>."
  • UPDATE: "That changes it: <new evidence> -> <new conclusion>."
  • CHECK: "Can't tell from here. Settling it: <command>." Then run it.

The receipt

ran: pytest -q
result: 118 passed, 2 skipped

or, when nothing ran:

unverified: no test covers the retry path

unverified: always passes the gate. "Done" with neither does not, in full and ultra.

When it gets in the way

/frank lite reports without interrupting. /frank off stops everything. Every hook fails open: on any internal error it goes quiet rather than blocking the turn.

More

https://github.com/HimanshuJ16/frank

Security

  • Security scanPassed, 95 of 100Pattern checks on the skill's files.
  • Model reviewPassed

See the full security report

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

See evaluation results

Source

Popularity

Repository stars
7
Repository forks
2
7-day trend score
10

Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/himanshuj16/frank/frank-help/badge)](https://verified-skill.com/skills/himanshuj16/frank/frank-help)

Something wrong with this skill or its page? Tell us.