Skip to content

Changelog

Entries apply to both @maestria/pi and @maestria/omp unless otherwise noted.

July 21-22, 2026

  • Pi: Sync workflow mode commands (fein/sonar/blitz) through canonical source pipeline.

July 21, 2026

Agent directives compressed to cut context usage by ~38% on the orchestrator prompt without removing rules or intent. Shared specialist rules now live once in global rules instead of being repeated per specialist.

July 14, 2026

Adventurer and architect outputs now require [verified] or [inferred] tags on every assumption.

New CRITICAL RULE mandates defined output contracts before delegating. Open-ended exploration must be scoped with time and resource limits.

The orchestrator now has a 5-trap pre-flight checklist to catch weak delegation prompts.

Delegation now prefers specifying what to achieve over how to achieve it.

Experiment framing, parallel speculation, incidental findings, first-principles decomposition

Section titled “Experiment framing, parallel speculation, incidental findings, first-principles decomposition”

Additional methodology improvements across all specialist agents.

July 14, 2026

COMMIT PROTOCOL step 7 now requires a numbered list of PR description sections: Summary, ## Changes, ## Testing, and ## Breaking Changes.

July 14, 2026

Aligned orchestrator push rules with Branch Discipline

Section titled “Aligned orchestrator push rules with Branch Discipline”

The question() call when landing on main/master is removed. The commit protocol now auto-checkouts a feature branch instead of asking.

Restructured from a 3-part narrative to a lean table-first format with change-type prefixes (+/~/-), breaking change markers (!), and test file annotations ((test)).

July 14, 2026

After every task, the agent must produce a ## Changes summary table instead of a free-form description.

The agent now checks internal docs, the user-facing docs site, the user-facing changelog, and changesets before committing.

July 11, 2026

Standard Pi/OMP skill mechanism for agent content

Section titled “Standard Pi/OMP skill mechanism for agent content”

The orchestrator dispatcher prompt and global agent rules now load as platform skills (.md files in skills/). The old custom file-loading approach is removed.

Specialist agents registered with subagent system

Section titled “Specialist agents registered with subagent system”

All 7 specialist agent types (adventurer, architect, builder, diagnose, planner, reviewer, writer) are now registered as proper agent types. Each specialist has the right tool permissions - read-only specialists (adventurer, architect, planner, reviewer, diagnose) can’t write files; builder and writer can.

July 10, 2026

The orchestrator can now dispatch parallel reviewers with different focus areas (Security, Architecture, Performance, UX, General). Issues are categorized [fix]/[dismiss]/[escalate].

Commit is fully autonomous - agent reads git log, composes correct conventional commit message, commits without asking. Push is automatic on feature branches, asks only on main/master.

Eliminate questions - autonomous philosophy

Section titled “Eliminate questions - autonomous philosophy”

Agents now exhaust data, document assumptions, and proceed instead of asking for permission. Mid-phase questions eliminated across all specialists.

  • Observation over reasoning - reviewer now prioritizes running code over logical argument.
  • Review triage pipeline - issues categorized and validated with conservative conflict resolution.

July 1, 2026

Agents now prefer code intelligence tools (MCP, CLI, or other) before falling back to grep and read when exploring the codebase.

June 27, 2026

Agents now check from first principles before adopting existing patterns, and check for existing solutions before building from scratch.

June 24, 2026

Project workflow configuration (.maestria/)

Section titled “Project workflow configuration (.maestria/)”

Ships the .maestria/ project workflow protocol. Projects using Maestria’s agent directives can now define .maestria/workflow.md for custom delegation sequencing and .maestria/rules.md for project-specific rules. The orchestrator loads these via @adventurer delegation.

Effort anthropomorphism guard and writing style guidance

Section titled “Effort anthropomorphism guard and writing style guidance”

The orchestrator and all specialists now include explicit guardrails against two common failure modes:

  • Don’t anthropomorphize effort - A new Critical Rule tells the dispatcher that thinking “that analysis would be too much work” is a category error for a machine that only routes. The same principle is instilled in all specialists via the global agent rules.

June 24, 2026

Single source of truth for agent directives

Section titled “Single source of truth for agent directives”

Agent prompts and global rules now live in one canonical location (packages/core/agent-directives/) and sync automatically to all platform plugins. The scripts/sync-all and scripts/check-sync tools handle the pipeline.

June 23-24, 2026 (Pi) / July 22, 2026 (OMP)

Initial release of @maestria/pi and @maestria/omp - portable extensions that bring Maestria’s spec-driven orchestration, workflow modes, and specialist subagents to the Pi and OMP coding agents.

  • 7 specialist agent types - adventurer, architect, builder, diagnose, planner, reviewer, writer with structured role definitions and tool permission boundaries
  • 3 workflow modes - fein (full pipeline with gates), sonar (research only), blitz (fast implementation), each routing to the appropriate subagent pipeline
  • Global rules injection - cross-cutting orchestration, delegation, and context-management rules injected into every session
  • Session compaction preservation - structured state summaries survive compaction events, keeping task tracking and work-in-progress intact across turns
  • Subagent dispatch - via @gotgenes/pi-subagents (Pi) or native task() tool (OMP), with 6-field handoff validation
  • /orchestrate, /review, /maestria-status commands - slash commands for dispatching work, requesting reviews, and inspecting extension state
  • Maker/checker enforcement - tool-level restriction prevents the orchestrator from implementing its own tasks
  • 4 methodology skills - orchestrator dispatcher, global agent rules, handoff contract, and iteration limits
  • Built-in task() dispatch - uses OMP’s native task() tool, no separate subagent package needed

June 22, 2026

Initial release of @maestria/pi - a portable Pi extension bringing spec-driven orchestration, workflow modes, and specialist subagents to the Pi coding agent.

  • 8 specialist prompt templates - one orchestrator plus seven specialist subagents
  • 3 workflow modes - fein, sonar, blitz
  • Global rules injection via before_agent_start event
  • Session compaction preservation
  • Subagent dispatch via @gotgenes/pi-subagents with 6-field handoff validation
  • /orchestrate, /review, /maestria-status commands
  • Maker/checker enforcement via tool-call interception
  • 2 methodology skills - handoff contract and iteration limits