Skip to content

v2.0.0 — Token Optimization Sprint

Choose a tag to compare

View all tags
@girishr girishr released this 26 Jun 08:08
· 306 commits to main since this release
v2.0.0
214b968

What's Changed

This release is the Token Optimization Sprint — a focused pass to reduce the size and noise of SpecPilot-generated files so AI agents spend fewer tokens on boilerplate and more on your actual project.

Changed

  • Terse AI agent mandates (CS-066): 8 critical mandates rewritten to compact style; tiered spec-first gate (Trivial/Feature/Architectural); static "Read in this order" list replaced with on-demand context routing table by task type.
  • Leaner project.yaml template (CS-067): entire rules: and ai_context: blocks removed; replaced with a single comment pointing to the AI agent config file.
  • Onboarding prompt in its own file (CS-068/CS-072): onboarding.md now holds the greenfield or brownfield prompt (selected during init/add-specs); printed to stdout after generation; prompts.md no longer contains onboarding sections.
  • docs.md removed (CS-069): was fully redundant — spec conventions enforced by validate, checklist duplicated mandate #5, CLI reference belongs in SpecPilot's own docs.
  • Conditional api.yaml (CS-070): new apiParadigm option (rest | cli | graphql | none); auto-inferred from framework; file skipped entirely for UI/mobile projects.
  • Slimmer prompts.md (CS-071): removed "Common Commands" and "AI Agent Guidelines" sections.
  • Terse spec file templates (CS-073): stripped prose, ## Cross-References footers, and verbose [TODO: ...] blocks from all generated templates.

Full test suite: 188/188 passing.

Assets 2
Loading