Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add commands/majiayu000/vibeguard/interviewgit clone --depth 1 https://github.com/majiayu000/vibeguardWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/commands/majiayu000/vibeguard/interview)<a href="https://agentmods.dev/commands/majiayu000/vibeguard/interview"><img src="https://agentmods.dev/badge/commands/majiayu000/vibeguard/interview.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00032 | $0.00723 |
| Opus 5 | $0.00016 | $0.00362 |
| Sonnet 5 | $0.00006 | $0.00145 |
| Haiku 4.5 | $0.00003 | $0.00072 |
Grade A, and why
VibeGuard: Interview scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
Core Concept (from Anthropic official best practices)
- Failure of large functions often stems from unclear requirements rather than incorrect implementation
- AI proactively interviews users to uncover unconsidered edge cases and technical trade-offs
- Output structured SPEC after interview, it is recommended to execute in new session to get clean context
Trigger condition
- Involves the development of new functions/modules
- Requirements description is vague or incomplete
- Changes affecting multiple modules (complexity routing 6+ file level)
Guardrails
- No code modifications, only analysis and interviews
- Up to 4 rounds of interviews, 2-4 questions each, keep the pace tight
- Don’t ask obvious questions, explore the difficulties that users didn’t expect
- When the user says "you decide", the recommended solution is given and recorded
Steps
-
Understand the initial requirements
- Read user requirement description ($ARGUMENTS or current context)
- Quickly scan relevant code and understand existing architecture and constraints
- Identify fuzzy points and undefined boundaries in requirements
-
In-Depth Interview (using AskUserQuestion tool)
Round 1: Functional Boundaries
- What are the core use cases? Which scenes are explicitly not allowed to be done?
- Are there any existing implementations with similar functionality that I can refer to?
Round 2: Technology Implementation
- Performance/latency requirements? Data storage preferences?
- What existing interfaces need to be compatible with?
Round 3: Boundary Cases
- How to deal with concurrency/race conditions? Expected behavior on error?
- Downgrade strategy when data volume is large?
Round 4: Acceptance Criteria
- How do you count it as "done"? What test coverage is required?
- Are there any performance benchmarks that must be passed?
Dynamically adjust the question direction based on the previous round of answers. If you have covered enough in previous rounds, you can end the interview early.
-
Generate SPEC
# Feature Spec: <function name> ## Overview One sentence description + core value ## Functional requirements - FR-01: ... ## Non-functional requirements - NFR-01: Performance/Safety/Compatibility Requirements ## Technical Design ### Scope of influence - New: ... - Revise: ... ### Interface definition ### Data model changes ## Boundary cases - EC-01: ... ## Acceptance criteria - [ ] AC-01: ... -
Confirm and save
- Show SPEC to user for confirmation
- Save to the project root directory
SPEC.md(or user-specified path) - Remind users: It is recommended to execute SPEC in a new session, clean context implementation is more reliable
Follow-up connection
- In a new session: read SPEC.md →
/vibeguard:preflightgenerate constraint set → implement by SPEC
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 5d ago First seen · 87 lines · 32 tokens per session scan A e7395ea3585e
VibeGuard: Interview is a command published in the GitHub repository majiayu000/vibeguard (41 stars, last pushed 2d ago), licensed MIT. It adds 32 tokens to every session and 723 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other commands, from other repositories
patterns
List makoto's integrity pattern catalog (read-only).
show
Read makoto's results ledger by key (read-only).
status
Show makoto's integrity-guardrail state (read-only).
spec
Use before writing code when requirements are ambiguous or a feature has non-obvious scope. Do NOT use for simple, self-evident changes.
claudemd-design-adopt
Generate a thin, fact-based DESIGN.md from a UI project's real design-token sources and wire it into project CLAUDE.md. Use when (1) the user asks to configure design specs / 配置设计规范 / 设计规范 for agents, (2) UI sessions keep inventing colors or spacing instead of using the project's existing tokens, (3) a UI project…
claudemd-refresh
Refresh the installed claudemd plugin to the latest released version in one shot (marketplace update → uninstall → install via the claude CLI). Use when the SessionStart banner reports a newer version, or /claudemd-doctor flags a stale plugin cache. Restart Claude Code afterwards — spec + manifest then sync…