discuss

A command that interviews you in five phases and saves a detailed description of a proposed feature or change. EARS is a structured way to write requirements so they can be checked clearly.

In plain words
What is it for?
Use it before planning a non-trivial feature or refactor; its saved requirements can connect the plan to later implementation and tests.
Why use it?
It turns an unclear idea into agreed requirements before planning begins, reducing missing details and misunderstandings.

Command

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 commands/fockus/skill-memory-bank/discuss
Clone the repo
git clone --depth 1 https://github.com/fockus/skill-memory-bank
Per session 19 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 5,661 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 $0.00019 $0.05661
Opus 5 $0.00010 $0.02831
Sonnet 5 $0.00004 $0.01132
Haiku 4.5 $0.00002 $0.00566

Measured yesterday against content hash d89ef08aec39, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, 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 yesterday.

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.

commands/discuss.md · 253 lines

How it starts

The opening of the file, as written. The whole thing — 253 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/mb discuss

Run a structured 5-phase interview that turns a fuzzy idea into an EARS-validated context/<topic>.md. The output feeds mb-traceability-gen.sh REQ → Plan → Test matrix and is read by /mb plan to link stages to requirements.

When to use

Before creating a non-trivial plan (/mb plan feature/refactor/...). Skip for trivial fixes — the overhead isn't justified.

Arguments

  • <topic> — short slug (kebab-case). Becomes the filename: .memory-bank/context/<topic>.md.

Workflow

Pre-flight

  1. Resolve the skill bundle root once and invoke every bundled helper through it:
SKILL_DIR="${MB_SKILLS_ROOT:-$HOME/.claude/skills/memory-bank}"   # memory-bank skill bundle root
[ -f "$SKILL_DIR/scripts/mb-interview-artifact-check.sh" ] || {
  echo "mb: skill bundle not found at $SKILL_DIR — set MB_SKILLS_ROOT" >&2; exit 2; }

This is the same resolution every other command file uses (commands/mb.md, commands/agree.md). Two ways of getting it wrong, both of which silently disable the mandatory close gate:

  • Never derive the root from $0. In an executable-Markdown snippet $0 is the shell, not this file, so $(dirname "$0")/.. resolves to the parent of the user's current directory — a path that either has no scripts/ (exit 127, gate never runs) or belongs to somebody else (a planted scripts/mb-interview-artifact-check.sh answers the gate with exit 0).
  • Never call a bundled script by a bare relative path (bash scripts/mb-…): the working directory during /mb discuss is the user's project, with the same two outcomes.

The existence check is part of the contract, not a nicety: an unresolvable bundle must stop the command loudly instead of letting a later helper call fail open.

  1. Resolve the active Memory Bank through $SKILL_DIR/scripts/_lib.sh::mb_resolve_path and use the resolved path as MB_PATH in every later step — the bank may be local (.memory-bank/), a global bank registered via /mb init --storage=global, or legacy. Never hardcode .memory-bank/: a project on a registered global bank has no local directory and would be refused although its bank is active. Refuse only when the resolver returns nothing (suggest /mb init).
  2. Compute CONTEXT_FILE = $MB_PATH/context/<topic>.md.
  3. If CONTEXT_FILE exists → ask AskUserQuestion: continue editing / overwrite / cancel.

Read the full file on GitHub · 253 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. yesterday First seen · 253 lines · 19 tokens per session scan A d89ef08aec39

Subscribe to this mod's changes

discuss is a command published in the GitHub repository fockus/skill-memory-bank (25 stars, last pushed 1mo ago), licensed MIT. It adds 19 tokens to every session and 5,661 once invoked, about $0.0001 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.