aiad-bridgeAIAD (AI-Augmented Development, ia-in-the-loop) bridge skill. Lets the human switch between AIAD (human-engine) and SDD (AI-engine) by treating an OpenSpec change as equivalent to a user story, via the command `aiad bridge`. In `to-sdd` it wraps a US as an OpenSpec change to delegate execution to SD
aiad-designAIAD (AI-Augmented Development, ia-in-the-loop) skill. Helps a human think before coding a user story, via the command `aiad design`. Two modes. In `explore` mode it diverges (a broad fan of options, including non-obvious ones, plus naming candidates) and then converges (contrasts them against expli
aiad-explainAIAD (AI-Augmented Development, ia-in-the-loop) skill. Explains code, libraries, patterns, errors, or technical concepts at the right level so the human learns, via the command `aiad explain`. It acts as a mentor that adapts the level (junior/mid/senior), explains the WHY and not just the WHAT, uses
aiad-journalAIAD (AI-Augmented Development, ia-in-the-loop) authorship-journal skill. Records and reports what the human authored versus what they delegated to the AI, via the command `aiad journal`. In `log` mode it appends a structured line to docs/aiad-journal.md (user story, skill, mode, who authored, one-l
aiad-pairAIAD (AI-Augmented Development, ia-in-the-loop) skill. Sustained turn-taking pair-programming session over a user story, via the command `aiad pair`. The human is the driver (writes the code) and the AI is the navigator (anticipates, watches the direction, suggests the next step, catches errors), wi
aiad-reviewAIAD (AI-Augmented Development, ia-in-the-loop) skill. Didactic review of the code the human wrote, via the command `aiad review`. Three focuses. `correctness` checks behavior, edge cases, errors, and coverage of the user story's acceptance criteria. `quality` checks readability, naming, duplication
aiad-rubber-duckAIAD (AI-Augmented Development, ia-in-the-loop) skill. Rubber-duck session to think out loud, via the command `aiad rubber-duck`. The human explains the problem and the AI acts as an active rubber duck: it listens, reflects back, asks Socratic questions, and helps them reach THEIR own answer. By def
aiad-saveAIAD (AI-Augmented Development, ia-in-the-loop) utility skill. Commits all pending changes and pushes the current branch WITHOUT asking any questions, via the command `aiad save`. A fast "save everything" for when you are in a hurry and just want your work uploaded. It stages everything (respecting
aiad-tddAIAD (AI-Augmented Development, ia-in-the-loop) skill. Human-first TDD support, via the command `aiad tdd`. The AI writes the failing tests for the target the human is about to build, and the human implements the production code until they go green. The target scope is the human's to set and can be
aiad-testAIAD (AI-Augmented Development, ia-in-the-loop) skill. Fills the missing tests over code the human has already written, via the command `aiad test`. Two modes. In `unit` mode it detects the repo's test framework and writes clear unit tests (positive, negative, edge) for the target the human names —
aiad-unblockAIAD (AI-Augmented Development, ia-in-the-loop) hub skill. Entry point for when the human is stuck and does not even know what kind of help they need, via the command `aiad unblock`. It does a quick triage of the blocker (approach, understanding, bug, quality, performance, tests, or engine) and rout