Keep Agent Engineering Decisions Current with Haft
Use Haft when an agent needs source-backed decision framing, comparison, authorization records, and stale-evidence checks across supported coding-agent hosts.
npx skills add agentskillexchange/skills --skill keep-agent-engineering-decisions-current-with-haft
Use Haft when an engineering agent needs to frame a decision, compare options under explicit parity rules, record the authorization behind a plan, or verify whether a previous decision is still supported by current evidence. The operator installs Haft, initializes it in a project, configures the chosen host adapter, and runs focused workflows such as /h-frame, /h-explore, /h-compare, /h-decide, /h-verify, and /h-status. Use this skill to run, configure, review, and manage project decisions that future agent work will depend on.
What this skill actually does
Invoke this instead of a normal note, a generic memory store, or an ad hoc prompt when later agent work must rely on a decision and needs replayable context: what was decided, what evidence supported it, whether the evidence has decayed, and whether a human decision or revised plan is now required.
The scope boundary is agent-assisted engineering decision governance inside a project. This is not a generic framework card, a broad CLI listing, or a replacement for the supported coding agents themselves. It is a local-first decision and project-memory layer for agents that need durable, source-backed reasoning records.