otaman

BA agent skill for user-flow and business-process registries — scaffolds flow and process drafts from natural-language prompts, audits completeness (missing required fields, dangling references, unreachable states), detects cross-artifact gaps (outcomes without flows, processes whose data-contracts are TBD), and guides HITL handoff to CTO review. Read-only on all live registries; draft output goes to staging. Use when the BA needs to author, review, or triage flows and processes.

CPO-agent skill for outcome management — scaffolds JTBD outcome drafts from natural-language prompts, audits completeness (missing as-a / i-want-to / incremental-outcome / so-i-can sub-fields), and surfaces inconsistencies (duplicate JTBD statements, P0s with stale Drafting status, orphaned solutions). Use when the CPO needs to author, review, or triage the program's outcomes.yaml.

Act as CTO and Pre-Sales Solution Architect for strategic and technical advisory at a software company across healthcare, fintech, e-commerce, AI/ML, gaming, drones/UAV, embedded/IoT, and other domains. Use when the user asks about technology decisions, architecture direction, team structure, vendor evaluation, compliance strategy, build-vs-buy, process improvement, or any executive-level judgment call on technical matters. Trigger on 'how should we...', 'which stack...', 'is it worth...', 'what's the right approach for...', even without 'as CTO'. Reads company-specific context from company-context.md if present and domain knowledge from references/domains/ for relevant domain(s). Otherwise gives context-appropriate generic advisory. Do NOT use for concrete project estimation with a brief, RFP, or budget/timeline output — that belongs to project-estimator. When the question is open-ended strategy, use this. When it's 'give me a number', use project-estimator.

Extracts reusable knowledge (facts, decisions, metrics, learnings) from artifacts processed during any phase

Core otaman skill — ownership enforcement, inter-agent communication protocol, and multi-repo coordination rules

Produce structured project estimates (proposal + workbook) for software delivery engagements across healthcare, fintech, e-commerce, AI/ML, gaming, drones/UAV, embedded/IoT, and other domains. Use when a user provides a project brief, client email, RFP, call notes, or technical requirements, or asks for an estimate, feasibility assessment, scoping, ballpark, budget range, or timeline. Trigger on 'estimate this', 'what would this cost', 'how long would X take', 'help me scope this', 'produce a proposal', 'ballpark this', or RFP responses. Also handles sales call notes converted into structured estimates. Deliverables follow a two-artifact model — a 5-slide proposal plus a 6-sheet workbook with specific schemas. Handles AI-assisted delivery calibration and domain-specific patterns. Reads company rates from company-context.md and domain knowledge from references/domains/. Do NOT use for open-ended strategy without a concrete project — that belongs to cto-advisor.

Observer skill for risk register and assumption registry review — audits completeness of risks.yaml and assumptions.yaml, flags stale entries (last-reviewed-at beyond program SLA), surfaces aggregate-score insights (top-N by score, exposure totals, category distribution), and suggests cross-references when new outcomes, solutions, or processes land without corresponding register updates. NO autonomous authoring — read-only observer only. Use when a human wants a structured health-check of the program's risk and assumption posture.

The Stage-2 Constitutional critic for the spec-proposal gate (spec-proposal-constitutional-gate, JTBD-57). Invoked ONLY inside a critic-session dispatch — a fresh Claude Code session belonging to a fleet agent whose own repo is NOT in the proposal's affected_repos, with zero shared context with the proposer. Applies the spec-critique-constitution rubric and emits a spec-proposal-critique-result bus message. Do not self-invoke on your own proposals or outside a dispatch.

Versioned rubric for the Stage-2 spec-proposal critic (spec-proposal-constitutional-gate, JTBD-57, design D3). NOT a user-facing skill — it is read by the spec-critique skill inside a fresh critic session and is never invoked directly. The critique SHALL cite this version.

OpenSpec delegation and spec change workflow — delegates to OpenSpec when present, provides lightweight fallback when not