decision

A guided record of an architecture decision: what was chosen, why it was needed, which alternatives were rejected, and when to reconsider it. Architecture means the major technical choices that shape a system.

In plain words
What is it for?
Use it to interview the team about one significant technical choice and record its context, alternatives, consequences, affected files or specifications, and a concrete signal for revisiting it.
Why use it?
It preserves the reasoning behind choices that may be costly to change or confusing to future maintainers. It also avoids filling the decision log with ordinary task notes.

Command for Claude Code

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/juanklagos/spec-driven-development-template/decision
Clone the repo
git clone --depth 1 https://github.com/juanklagos/spec-driven-development-template

Made for: Claude Code.

Per session 35 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 790 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 $0.00035 $0.00790
Opus 5 $0.00017 $0.00395
Sonnet 5 $0.00007 $0.00158
Haiku 4.5 $0.00003 $0.00079

Measured 2d ago against content hash 7f8686a28c42, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

decision 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 2d 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.

.claude/commands/sdd/decision.md · 35 lines

What it actually says

Capture ONE decision and write it to the decision log. Respond in the user's language (EN/ES). Ask one short question at a time; never dump the whole questionnaire at once.

  1. Check it is worth recording. A decision belongs in the log when at least one is true:

    • it chose between real alternatives (something else was actually on the table),
    • it will be expensive to reverse (multiple specs, migration, contract or license change),
    • a future reader would ask "why is it like this?" and the code alone would not answer. If none holds, say so and stop — a task note in history.md is enough. Do not pad the log.
  2. Interview the user (skip anything $ARGUMENTS already answers):

    • What was decided — one sentence, in the imperative or the past tense, no hedging.
    • Context — what problem forced the choice? what was true at that moment?
    • Alternatives — what else was really considered, and why each was rejected. "None" is a valid answer; write "no alternatives were considered" rather than inventing a straw man.
    • Consequences — what improves, what trade-off is accepted, which specs/files are affected.
    • When to revisit — the concrete signal that should reopen this decision (a benchmark, a date, a dependency, a limit being hit). Summarize back in 5 bullets and get a "yes" before writing.
  3. Ground every claim in evidence. Each rationale MUST point at a source: a commit (hash + date from git log, never a guessed date), a file:line, a spec history.md entry, a CHANGELOG.md line, a document in idea/, or the user's own words in this session. If a claim cannot be sourced, write it down as unsourced ("no written rationale exists; reconstructed from the diff") instead of inventing one. A false decision record is worse than no record.

  4. Write the file:

    • Path: bitacora/decisiones/YYYY-MM-DD-<slug>.md (sidecar projects: ./spec/bitacora/decisiones/...).
    • Date: today's real date (date +%F), or the date of the commit the decision actually landed in if you are recording it after the fact.
    • Slug: short kebab-case, in the user's language, describing the decision — not the feature (no-desktop-app, not desktop).
    • Structure: copy bitacora/templates/DECISION_TEMPLATE.md and fill every section, including When to revisit. Keep the bilingual headings.
  5. Link it back: mention the record in the active spec's history.md and in bitacora/global/PROJECT_LOG.md, and reference related records by filename when a decision extends, narrows or supersedes an earlier one. If it supersedes one, say so in both files — the old record stays, it is not deleted.

  6. Confirm to the user with the path written and a one-line summary, and offer to add it to bitacora/decisiones/README.md.

Never fabricate a rationale, a date, or an alternative that was not really on the table. The value of this log is that it is true. / Nunca inventes justificación, fecha ni alternativa: el valor de esta bitácora es que es verdadera.

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. 2d ago First seen · 35 lines · 35 tokens per session scan A 7f8686a28c42

Subscribe to this mod's changes

decision is a command published in the GitHub repository juanklagos/spec-driven-development-template (15 stars, last pushed 2d ago), licensed MIT. It adds 35 tokens to every session and 790 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.