Architecture decision records
Skill ldilov/harness-forge/skills/architecture-decision-records
AI coding agent runtime that makes LLMs understand your codebase with repository-aware context, memory, and continuous learning
npx -y skills add ldilov/harness-forge --skill architecture-decision-recordsAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 3 stars3 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
create and maintain high-signal adrs for architecture, package, runtime, workflow, and operational decisions. use when a decision should stay reviewable after the current session or when multiple credible options need a durable rationale.
SKILL.md
2.4 KB, as published. Nobody here has run it
Architecture Decision Records
Trigger Signals
- the task changes architecture, package boundaries, runtime behavior, support guarantees, or operating model
- more than one credible option exists and the trade-off should stay reviewable later
- a decision will affect multiple teams, releases, or future migrations
Inspect First
- the affected specs, tasks, plans, and runtime boundaries
- any existing ADRs, design docs, or release notes that already capture related decisions
- operational constraints such as support window, migration cost, reliability, security, or compliance
Workflow
- frame the actual decision, not the implementation diary around it
- capture the strongest drivers, constraints, and alternatives that were seriously considered
- choose a concise ADR shape and write context, decision, status, and consequences in plain language
- note follow-up actions, migration implications, and what would invalidate the decision later
- update or supersede older ADRs explicitly when the decision changes an earlier one
Output Contract
- ADR title, status, and date-ready decision summary
- short context section with the forces that matter
- options considered and why the chosen path won
- consequences, follow-up actions, and superseded links when relevant
Failure Modes
- the real decision has not been framed clearly yet
- the work is still exploratory enough that no stable rationale exists
- the note drifts into a long implementation log instead of a decision record
Escalation
- escalate when the decision changes public contracts, support commitments, or cross-team ownership
- escalate when alternatives have materially different operational or migration risk
- escalate when the ADR appears to conflict with an existing record and the supersession path is unclear
References
skills/architecture-decision-records/references/adr-template.mdskills/architecture-decision-records/references/decision-rubric.mdskills/architecture-decision-records/references/madr-style-guide.mdskills/architecture-decision-records/references/adr-anti-patterns.mdskills/architecture-decision-records/references/examples.md