Skip to content

Changelog

October 5, 2026

  • Requires Node.js 22.19 or later.

September 30, 2026

  • Agents reassess explanatory comments and redundant tests before accepting changes, and shared directive text was consolidated with no behavior change; see the methodology changelog.

September 25, 2026

  • Agents follow behavior-first testing guidance and deliver work as verified, reviewable slices with designs shown before implementation; see the methodology changelog.

September 24, 2026

  • Project customization (.maestria/workflow.md, .maestria/rules.md) now loads through one shared loader across integrations; ordering, live reloads, and error reporting are unchanged.

September 23, 2026

  • Project customization (.maestria/workflow.md then .maestria/rules.md) loads as subordinate guidance; see the methodology changelog.
  • create-pull-request and docs-update skills install through the skills CLI with per-skill selection; see the CLI changelog.

September 18, 2026

  • Consistent PR delivery contract with proportional documentation assessment; see the methodology changelog.

September 17, 2026

  • Stricter visual delivery evidence contract, with capture, PR-body publication, and readback as distinct stages; see the methodology changelog.

September 16, 2026

  • Defined when Sonar research is complete, with Blitz reusing valid context and switching routes for investigation or design work; see the methodology changelog.

September 15, 2026

  • Clarified rendered verification and visual delivery evidence, keeping required artifacts and verification gaps in handoffs; see the methodology changelog.

September 15, 2026

  • Internal consolidation; shared the review-model context type through the shared review core and delegated mode keywords to the shared-mode package. No user-facing behavior change; generated agent and skill output is unchanged.

September 11, 2026

  • Internal consolidation; shared more host seams with the private shared packages and removed dead adapters. No user-facing behavior change; generated agent and skill output is unchanged.

September 8, 2026

  • Stopped publishing generated JavaScript sourcemaps to reduce published package sizes; compiled runtime artifacts are unchanged.

September 6, 2026

  • Clarified installation paths, prerequisites, and support boundaries in the package documentation.
  • Cross-cutting agent methodology update; see the methodology changelog.

September 4, 2026

September 3, 2026

August 25, 2026

August 24, 2026

August 22, 2026

August 14, 2026

August 13, 2026

Initial release

@maestria/prime-agent delivers the maestria engineering methodology for Prime Agent as standard Agent Skills - 7 specialist roles, orchestrator, global-rules, handoff and iteration-limits aids, and fein/sonar/blitz workflow modes - generated from the canonical directives by the root sync pipeline (skills/<name>/SKILL.md with required name/description frontmatter) - plus a small, verified Prime/Pi extension (dist/extension.mjs, pi.extensions) covering workflow-mode slash commands (/fein, /sonar, /blitz, /mode-clear, /maestria-status) and before_agent_start mode prompt injection with session-scoped mode state (custom session entries, no filesystem writes).

  • Native candidate, verified subset only. The extension uses only the public extension API of the pinned Prime fork (commit 7787f074...), performs no tool interception, and has no runtime dependency on pi packages. Evidence re-verified on 2026-08-13 (see ADR-CORE-014 and the runtime support matrix).
  • Native recursive-subagent (rlm) dispatch and JSON/RPC headless mode remain deferred - the pinned fork exposes no public JS extension bridge for rlm.
  • No sandbox or runtime-enforcement claim is made.
  • Reuse the npm pack dry-run result across package assertions so the packaging tests do not exceed Vitest’s per-test timeout.