discuss

discuss is a skill for Claude Code, Codex from shawnla90/recursive-drift. It costs 56 tokens per session (1,287 once invoked), scanned A, original, MIT.

A decision-lock workflow for planning complex coding features before implementation. It asks focused questions and writes a CONTEXT.md file that later work can follow.

In plain words
What is it for?
Use it for changes spanning at least three files and multiple concerns, such as interface, data and operations work. It can also be started manually with /discuss.
Why use it?
It reduces repeated questions and prevents coding from starting before important scope and design choices are settled.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one.

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/shawnla90/recursive-drift/discuss
Any agent
npx skills add shawnla90/recursive-drift --skill discuss
Clone the repo
git clone --depth 1 https://github.com/shawnla90/recursive-drift

Made for: Claude Code, Codex.

Wrote 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.

agentmods badge for discuss

README.md
[![agentmods](https://agentmods.dev/badge/skills/shawnla90/recursive-drift/discuss.svg)](https://agentmods.dev/skills/shawnla90/recursive-drift/discuss)
Your own site
<a href="https://agentmods.dev/skills/shawnla90/recursive-drift/discuss"><img src="https://agentmods.dev/badge/skills/shawnla90/recursive-drift/discuss.svg" alt="Measured on agentmods" height="20"></a>
Per session 56 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,287 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 5.1 $0.00056 $0.01287
Opus 5 $0.00028 $0.00643
Sonnet 5 $0.00011 $0.00257
Haiku 4.5 $0.00006 $0.00129

Measured 6d ago against content hash f506b2572686, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-05, from the pricing page.

Security

Grade A, and why

discuss 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 6d 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.

skills/discuss/SKILL.md · 138 lines

How it starts

The opening of the file, as written. The whole thing — 138 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Discuss -- Decision-Lock Before Execution

Lock all decisions BEFORE any code gets written. One CONTEXT.md per feature. No agent re-asks what's already decided.


When to Invoke

  • Auto: agent-routing Step 0 detects 3+ files across 2+ concerns and no CONTEXT.md exists
  • Manual: User types /discuss or "let's discuss this first"
  • Skip: Task is 1-2 files in a single concern, or CONTEXT.md already exists for this feature

Step 1: Domain Detection

Classify the work into one or more domains. Each domain has its own question bank -- never ask generic questions.

Domain Signals Question Focus
Visual/UI Components, pages, layouts, CSS, animations Responsive breakpoints, existing component reuse, design system tokens, interaction states
API/Data Endpoints, data files, types, schemas, database Data shape, validation rules, error handling, existing type patterns
Operations Crons, deploy, CI, scripts, infra Schedule, failure handling, monitoring, rollback strategy
Content Blog posts, wiki entries, social, copy Voice compliance, platform, audience segment
Organizational Skills, constraints, workflows, team config Scope of impact, backward compatibility, which skills consume the output

Multi-domain tasks get questions from each relevant domain.


Step 2: Scope Boundary Lock

State the scope explicitly in 2-3 sentences:

Scope: We are building [X]. This includes [A, B, C]. This does NOT include [D, E].

Present to user. They confirm with "yes", narrow with "remove X", or expand with "also include Y."

Anything the user mentions that falls outside the stated scope -- add to Deferred Ideas (Step 4).


Step 3: Ask 5-7 Targeted Questions

Pull from the domain question banks detected in Step 1. Rules:

  • Max 7 questions. If you need more, the scope is too broad -- split into two sessions.
  • Reference existing code by path. Not "what pattern?" but "should this follow the pattern in path/to/existing/file or a new layout?"
  • Offer recommended defaults. Every question should have a (recommended: X) when the agent has an informed opinion.
  • No vibes. Every answer must be a concrete, implementable decision. "Make it look good" is not a decision. "Use the existing grid component with 3 columns on desktop, 1 on mobile" is.
  • Ask all questions at once. Do not drip-feed. Present the full batch and let the user answer in one pass.

Read the full file on GitHub · 138 lines

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. 6d ago First seen · 138 lines · 56 tokens per session scan A f506b2572686

Subscribe to this mod's changes

discuss is a skill published in the GitHub repository shawnla90/recursive-drift (18 stars, last pushed 5mo ago), licensed MIT. It adds 56 tokens to every session and 1,287 once invoked, about $0.0003 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