timdevai decision-recorder
When you say "we decided to go with X over Y," this skill writes an ADR (Architecture Decision Record) so you don't have to remember why six months later.
When to activate
Trigger phrases: we decided, go with, chose, picked X over Y, let's settle on, final answer.
Listen passively during planning conversations. When a real decision is reached, surface: "Save this as an ADR?"
Output template (Michael Nygard format, lightly modified)
# ADR-NNN: <Decision title>
Date: <YYYY-MM-DD>
Status: Accepted | Superseded by ADR-NNN | Deprecated
## Context
<What forces led to this decision? 2-4 sentences.>
## Decision
<One-paragraph statement of what we're doing.>
## Alternatives considered
- <Option A>: <why not>
- <Option B>: <why not>
- <Option C>: <why not — if any>
## Consequences
### Positive
- <good thing>
- <good thing>
### Negative
- <real tradeoff>
- <real tradeoff>
### Neutral
- <thing that just is>
Numbering
- Find existing
docs/adr/(ordocs/decisions/,architecture/decisions/) - Auto-number the next file: ADR-0001, ADR-0002, etc.
- Filename:
docs/adr/ADR-NNNN-<slug>.md
Process
- Detect decision moment from conversation
- Extract: context, decision, alternatives mentioned, tradeoffs
- Generate draft ADR
- Ask "Save as ADR-NNNN-.md?"
- On approval, write the file
- Suggest committing it via
commit-curator
Inputs
- Conversation context where decision was made
- Existing
docs/adr/(for numbering)
Outputs
- New ADR file
- ADR number used
Slash command
/adr [decision summary] — manually trigger an ADR for a decision the model may have missed.
Example
Conversation:
User: should we use Postgres or DynamoDB? Claude: ... User: ok let's go with Postgres, the relational queries we need outweigh the scaling concerns Skill: Detected decision. Drafting ADR-0007: "Use Postgres as primary datastore." (shows draft) User: y Skill: Wrote docs/adr/ADR-0007-postgres-primary-datastore.md