This roadmap grows AgencyOps in response to real work. It is a sequence, not a promise to create every possible agency document.
- Work on the current phase before expanding the next one.
- Add a document only when it will be used in an active or immediately upcoming workflow.
- Limit each change to one operational problem when practical.
- At the end of each phase, remove duplication and defer anything without a clear owner.
Goal: establish the repository contract without inventing processes.
- Define scope, structure, operational knowledge model, document lifecycle, and naming conventions.
- Add contributor guidance for humans and AI assistants.
- Establish a staged roadmap.
- Choose initial maintainers and the pull-request approval rule.
- Add lightweight Markdown link and style checks when the first operational document is introduced.
Exit condition: contributors can decide what belongs here, where it belongs, and how it should be changed.
Goal: describe the minimum shared path and quality bar for delivering one web project.
- Create
standards/definition-of-done.mdfrom a completed project's actual acceptance checks. - Create
sops/project-lifecycle.mdwith phases, owners, entry conditions, and exit conditions. - Create
templates/project-brief.mdonly after the lifecycle identifies its required inputs. - Adopt and test
templates/PROJECT.mdas the living operational record on one project. - Use all three on one project and record friction as issues rather than immediately adding more documents.
- Review after the pilot; merge overlap and remove steps that were not useful.
Exit condition: one project has used the baseline from intake through handoff, and the documents reflect what actually worked.
Goal: make common quality expectations and project changes predictable.
- Create
standards/web-baseline.mdwith measurable accessibility, performance, SEO, security, and compatibility expectations. - Map automatable requirements to checks used by starters and projects.
- Create
sops/change-control.mdfrom real scope-change and approval needs. - Add no more than two supporting templates, and only where repeated manual recreation is evident.
- Pilot the standards on a second project and resolve conflicts with the Phase 1 baseline.
Exit condition: two projects can apply the same quality baseline and handle a change without relying on undocumented knowledge.
Goal: connect policy to executable starter repositories without coupling their source histories.
- Create
starters/astro-contract.mddefining required capabilities, checks, metadata, and allowed deviations. - Define the minimal
agencyops.jsonschema and include the adopted AgencyOps release or commit. - Adapt one Astro starter as the reference consumer.
- Verify that canonical prose remains in AgencyOps while enforcement remains in the starter.
- Document the upgrade procedure using one real AgencyOps baseline change.
Exit condition: a starter declares its AgencyOps baseline, enforces objective requirements, and can be upgraded through a reviewed process.
Goal: add guidance only in areas where repeated work or mistakes justify it.
Choose the next item from evidence gathered in projects; do not pursue all tracks at once.
- Add the highest-value missing process, likely discovery, launch, handoff, support, or incident response.
- Write a playbook when a recurring situation needs judgment beyond a linear process.
- Record the first consequential repository decision as
decisions/0001-short-title.md. - Add an ownership review cadence and review current
standarddocuments. - Evaluate search, generated indexes, or shared packages only after repository size makes navigation or synchronization measurably difficult.
Exit condition: every added artifact has an owner, evidence of use, and no competing canonical document.
Possible future capabilities include automated metadata validation, generated documentation indexes, a versioned shared configuration package, structured onboarding paths, and a counsel-approved legal-template library with explicit governance. Adopt them only when the maintenance cost is lower than the demonstrated cost of the problem they solve.