Skip to content

Team Roster

Generated from .dacli/ by dacli catalog — do not edit this page. It is a one-way read view: to change a role or skill, edit its file under .dacli/ (via PR), then regenerate. Versions and last-changed come from git history.

A role's Grant must agree with its runtime. A ro grant is only honest on a runtime that can enforce read-only, so dacli spawn --grant ro on a runtime with no read-only sandbox is refused (exit 3), never downgraded to rw — check it with dacli runtime doctor (a runtime shown ✗ no read-only mode cannot back a ro role). An rw grant is also refused when the runtime's allowlist grants no write tool (Edit or Write); a runtime with no allowlist makes no such promise and is treated as writable. --cooperative explicitly overrides either capability refusal. Runtime and model routing are shown below; the adapter allowlist is in .dacli/runtimes/<name>.md.

Roles (18)

Role Version Grant Kind Runtime Model Tier / capacity / context Skills Purpose Last changed
adversarial-reviewer v1 rw reviewer codex-rw gpt-5.6-sol tier 3 / ≤12 pt / 200k ctx using-dacli, evidence-verification, go-system-design, runtime-process-safety, github-delivery provide cross-runtime adversarial review when provider independence matters; cooperative read discipline, never unattended —
estimator v2 rw planner codex-rw gpt-5.6-terra tier 1 / ≤3 pt / 200k ctx using-dacli, evidence-verification, product-research-design PM-style role: sizes an open task with a three-point estimate derived from the codebase, not typed 10 days ago · chore: move this repo's workspace off trunk onto the dacli-record ref (#393)
fixer v4 rw implementer codex-rw gpt-5.6-terra tier 1 / ≤8 pt / 200k ctx using-dacli, evidence-verification, go-system-design, github-delivery implement one scoped task end to end in Go — failing test first, smallest change, then land it 10 days ago · chore: move this repo's workspace off trunk onto the dacli-record ref (#393)
frontend-engineer v2 rw implementer codex-rw gpt-5.6-terra tier 2 / ≤8 pt / 200k ctx using-dacli, evidence-verification, frontend-quality, github-delivery build the Vue dashboard SPA — TypeScript, Composition API, Pinia, Vite, component-driven, accessible 10 days ago · chore: move this repo's workspace off trunk onto the dacli-record ref (#393)
frontend-reviewer v2 ro reviewer cc sonnet tier 2 / ≤6 pt / 200k ctx using-dacli, evidence-verification, frontend-quality review Vue/TS frontend for best practices, accessibility, performance, and design fidelity; never implements 10 days ago · chore: move this repo's workspace off trunk onto the dacli-record ref (#393)
go-auditor v4 ro reviewer cc opus tier 3 / ≤8 pt / 200k ctx using-dacli, evidence-verification, go-system-design, runtime-process-safety audit Go correctness, persistence, concurrency, and measured performance without implementing 10 days ago · chore: move this repo's workspace off trunk onto the dacli-record ref (#393)
integrator v4 rw reviewer codex-rw gpt-5.6-sol tier 3 / ≤6 pt / 200k ctx using-dacli, evidence-verification, github-delivery, runtime-process-safety merge done-task PRs to trunk on green CI, autonomously — never implements 10 days ago · chore: move this repo's workspace off trunk onto the dacli-record ref (#393)
junior v3 rw implementer codex-rw gpt-5.6-terra tier 1 / ≤3 pt / 200k ctx using-dacli, evidence-verification, github-delivery small well-scoped tasks on the cheap model 10 days ago · chore: move this repo's workspace off trunk onto the dacli-record ref (#393)
loop-auditor v1 ro reviewer cc opus tier 3 / ≤12 pt / 200k ctx using-dacli, evidence-verification, go-system-design, runtime-process-safety, github-delivery audit one completed loop wave, file only evidence-backed non-duplicate work, and never edit product code —
maintainer v4 rw implementer codex-rw gpt-5.6-sol tier 3 / ≤13 pt / 200k ctx using-dacli, evidence-verification, go-system-design, runtime-process-safety, github-delivery implement high-blast-radius dacli changes across architecture, persistence, runtimes, and delivery 10 days ago · chore: move this repo's workspace off trunk onto the dacli-record ref (#393)
mutation-auditor v2 ro reviewer cc opus tier 3 / ≤8 pt / 200k ctx using-dacli, evidence-verification, go-system-design prove the test suite actually measures what it claims: break the code a passing test covers and confirm the test fails —
prompt-auditor v1 ro reviewer cc sonnet tier 2 / ≤8 pt / 200k ctx using-dacli, evidence-verification, runtime-process-safety audit and sharpen the agent prompt registry 10 days ago · chore: move this repo's workspace off trunk onto the dacli-record ref (#393)
reviewer v3 ro reviewer cc opus tier 3 / ≤13 pt / 200k ctx using-dacli, evidence-verification, go-system-design, runtime-process-safety, github-delivery independently review high-consequence changes for correctness, evidence, and contract drift; never implements 10 days ago · chore: move this repo's workspace off trunk onto the dacli-record ref (#393)
role-architect v1 rw designer codex-rw gpt-5.6-sol tier 3 / ≤8 pt / 200k ctx using-dacli, evidence-verification, runtime-process-safety, product-research-design provision the minimal roster an adopted codebase actually needs — each role justified by code that exists, with method written, not metadata 10 days ago · chore: move this repo's workspace off trunk onto the dacli-record ref (#393)
seam-auditor v2 ro reviewer cc opus tier 3 / ≤8 pt / 200k ctx using-dacli, evidence-verification, go-system-design, runtime-process-safety, github-delivery audit COMPOSITIONS of individually-correct features, where this codebase's expensive bugs actually live —
ui-ux-designer v2 rw designer codex-rw gpt-5.6-terra tier 2 / ≤6 pt / 200k ctx using-dacli, evidence-verification, frontend-quality, product-research-design own the dashboard's UX and visual design — layout, information hierarchy, the mission-control aesthetic, responsive + a11y; produce component/design specs the engineer builds to 10 days ago · chore: move this repo's workspace off trunk onto the dacli-record ref (#393)
ux-researcher v2 rw researcher codex-rw gpt-5.6-terra tier 1 / ≤8 pt / 200k ctx using-dacli, evidence-verification, product-research-design plan and run dashboard UX + product-discovery research: interview guides, synthesis, personas, opportunity framing, and a prioritized feature roadmap grounded in evidence 10 days ago · chore: move this repo's workspace off trunk onto the dacli-record ref (#393)
visionary v1 ro researcher cc sonnet tier 2 / ≤8 pt / 200k ctx using-dacli, evidence-verification, product-research-design research and upgrade the product vision, features, and direction 10 days ago · chore: move this repo's workspace off trunk onto the dacli-record ref (#393)

Skills (7)

Skill Version Est. tokens Purpose Last changed
evidence-verification v1 34 Use when implementing, reviewing, or accepting changes that need reproducible evidence, mutation testing, and honest verification records. —
frontend-quality v2 28 Use for the Vue dashboard, TypeScript, accessibility, responsive design, state coverage, and frontend performance. —
github-delivery v1 20 Use for GitHub-first issues, PRs, checks, idempotent synchronization, and landing. —
go-system-design v2 29 Use for Go architecture, persistence, concurrency, filesystem state, command contracts, and performance work in dacli. —
product-research-design v2 27 Use for product discovery, UX research, personas, prioritization, specifications, and architecture proposals. —
runtime-process-safety v2 28 Use for coding-CLI adapters, process trees, worktrees, grants, model routing, usage, cancellation, and recovery. —
using-dacli v4 39 How to work inside a dacli workspace: the task contract, exit codes, and how to record work so it counts. Triggers whenever you are spawned onto a dacli task. 10 days ago · chore: move this repo's workspace off trunk onto the dacli-record ref (#393)