Nedal7707
UserEvidence-backed guardrails for AI coding agents. Vibe fast. Ship with proof.
Categories
Indexed Skills (12)
complete-workflow
Use in EVERY session before substantive work. Enforces the WorktreeProof L99 complete workflow — goal/contract, plan, reserve, run, evidence, review, merge, release — with the Chrome/Computer/WorktreeProof plugins and the curated skill stack. Triggers on any task, session start, multi-step work, release, or launch.
best-practice-guard
Apply stack-aware security, accessibility, compatibility, performance, testing, and maintainability gates using current sources and measured evidence. Use when reviewing a change, selecting quality checks, or deciding whether a broad practice claim is justified.
resource-efficient-coding
Choose bounded CPU, RAM, disk, concurrency, cache, and artifact practices from measured host resources. Use when planning or reviewing parallel coding work, CI jobs, builds, cleanup inventories, or evidence collection where resource pressure and safe recovery matter.
token-efficient-context
Prepare small, fresh task briefs from only the rules, specs, sources, tests, and errors that matter, with explicit budgets and stop conditions. Use when context is large, stale, conflicting, sensitive, or expensive to load.
tool-orchestrator
Discover and choose currently available tools by capability, risk, and evidence needs. Use when a task spans local commands, connectors, plugins, browsers, or external services and tool access or side effects must be made explicit.
ui-proof-loop
Verify UI screens, states, and actions in short evidence rounds with redaction and operator review. Use when a change needs browser or desktop interaction proof, a visual handoff, or a careful boundary around sensitive UI data.
vibe-to-verified
Turn an early product idea into a bounded, testable, previewable change with rollback and closure evidence. Use when a non-expert needs help moving from a rough request to scoped lanes, checks, a safe preview, and a terminal handoff.
worktree-proof-stack
Move bounded coding-agent work from an initial idea to verified terminal closure with target checks, non-overlapping scopes, safe execution, and redacted evidence. Use when planning, delegating, running, integrating, or closing work that must remain auditable.
omnibus-maintainer
Route common open-source maintenance work through a concise, progressive-disclosure workflow. Use when triaging issues, improving tests or docs, preparing releases, reviewing dependencies, or responding to routine maintenance requests.
protocol-client
Negotiate a public WorktreeProof manifest and complete a bounded local workflow without forwarding private context. Use when a coding agent needs portable capabilities, relative scope, preview-only client guidance, or reversible local migration.
safe-parallel-delegation
Design safe, model-agnostic parallel work with unique lane identifiers, non-overlapping scopes, and terminal receipts. Use when splitting coding, review, test, or documentation work across independent execution surfaces.
worktree-proof
Coordinate bounded coding-agent lanes with explicit scopes, reservations, run records, and closure receipts. Use when parallel work needs conflict checks, auditable handoffs, stale-lane cleanup, or deterministic validation.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.