interview-framework-codex

An internal set of rules for conducting interviews during Ralph Specum planning phases. It governs discovery, user approval, skill selection, and when work may be handed to another agent.

In plain words
What is it for?
Use it internally when coordinating the start, triage, research, requirements, design, or task-planning phases.
Why use it?
It keeps phase decisions traceable and prevents delegation before the necessary context and approvals are complete. It is intended for the workflow coordinator rather than direct user use.

Skill for Claude CodeCodex

Install

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.

agentmods
npx agentmods add skills/tzachbon/smart-ralph/interview-framework-codex
Any agent
npx skills add tzachbon/smart-ralph --skill interview-framework-codex
Clone the repo
git clone --depth 1 https://github.com/tzachbon/smart-ralph

Made for: Claude Code, Codex.

Per session 45 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 664 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00045 $0.00664
Opus 5 $0.00023 $0.00332
Sonnet 5 $0.00009 $0.00133
Haiku 4.5 $0.00005 $0.00066

Measured 3d ago against content hash 7683dda900c1, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

interview-framework-codex 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 3d 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.

plugins/ralph-specum-codex/skills/interview-framework-codex/SKILL.md · 41 lines

How it starts

The opening of the file, as written. The whole thing — 41 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Codex interview framework

Run every normal-mode start, triage, research, requirements, design, and tasks phase through references/algorithm.md and references/domain-modeling.md.

Completion criterion

Delegate only after all of these conditions hold for one identity tuple:

  • phase, interviewId, discoveryRevision, and contextDigest match in state and in the delegation packet.
  • The selected skill manifest has status complete or partial_warned.
  • The phase interview has status complete or skipped after explicit final approval.
  • phase_gate.py check-delegation succeeds.

Quick mode uses the same identity tuple. It requires a current complete or partial_warned manifest plus an interview receipt with status bypassed_quick and quickAuthorization.source: "--quick". Quick mode bypasses interview questions only. Discovery, contract load, parent delegation provenance, artifact-agent reload, and check-agent-write still apply.

Hard boundaries

  • Ask only critical user decisions that can change phase scope, externally visible behavior, acceptance, architecture, sequencing, or material risk.
  • Inspect code, configuration, state, prior artifacts, and selected skill contracts for facts. Do not ask the user for discoverable facts or setup and administration choices.
  • Ask the whole currently unblocked critical frontier. Split a frontier only when Codex's native user-input tool limit requires another batch. The current limit is three questions per call.
  • Put the recommended option first and state its tradeoff. Offer only viable alternatives.
  • Persist each partial answer before asking the next frontier.
  • Treat control-only replies such as apply the changes, continue, proceed, and go ahead as no answer to an active interview question.
  • Treat bare skip during an active question as an instruction to finish the remaining interview with stated defaults and assumptions. Present the final approval gate before delegation.
  • Delegate in the same turn after the user explicitly chooses approve and delegate.
  • During artifact review, treat apply the changes as a revision request. Delegate the revision and remain at the artifact approval gate.
  • Apply domain-language modeling during the grill. Challenge conflicting or vague terms, use concrete boundary scenarios, and record resolved domain terms without creating an ADR.

Read the full file on GitHub · 41 lines

Files

What ships with it

3 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

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.

  1. 3d ago First seen · 41 lines · 45 tokens per session scan A 7683dda900c1

Subscribe to this mod's changes

interview-framework-codex is a skill published in the GitHub repository tzachbon/smart-ralph (527 stars, last pushed 5d ago), licensed MIT. It adds 45 tokens to every session and 664 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.

Related

Other skills, from other repositories