MVPavan/via / .claude/skills/execution
execution
VerifiedUse to execute tracked multi-step work, a standalone plan, or a roadmap phase. Clear bounded implementation can proceed directly under AGENTS.md.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 4.5 KB |
| agents/openai.yaml | config | 42 B |
| references/task-engine.md | markdown | 6.0 KB |
| references/workstream-mode.md | markdown | 2.1 KB |
| scripts/review-package.shexecutable | script | 2.7 KB |
| scripts/task-brief.shexecutable | script | 1.6 KB |
references/workstream-mode.md
View on GitHub# Workstream mode — unattended multi-phase execution
Enter only through an explicit request to run the remaining phases or `/run-phases`.
Reading this reference is not authorization. Record the authorized roadmap, phase
scope, plan approval coverage and commit policy in the execution ledger.
## Preflight
Capture the starting revision and dirty-file baseline. Reconcile each roadmap
phase ID with exactly one Beads epic title `[<phase-id>]`; report missing or
ambiguous records before execution. Preserve all pre-existing edits.
## Walk
1. Read roadmap order and Beads status/dependencies. Select the first eligible
unfinished phase; never cross an unmet prerequisite.
2. Execute it through the execution skill's phase loop and close gate. Unattended
authority covers routine plan elaboration within the agreed roadmap, not new
behavior, changed safety boundaries or scope expansion.
3. After the stage gate and phase exit pass, regenerate tracking and refresh the
Beads export where the authorized write scope permits it.
4. Honor the mode's established per-phase commit convention only when the user's
invocation covers it. An explicit no-commit restriction overrides it. Stage
named phase-owned paths only; if they include pre-existing edits, prepare a
scoped result rather than committing unrelated content. No push is implied.
5. Continue eligible phases until the authorized scope is complete. If work is
unfinished but no phase is eligible, report the blocking dependencies.
## Context and recovery
Stay in the current context while useful. Compact or hand off only when the
runtime supports it and context pressure warrants it; no unconditional slash
command applies across providers. Before recovery, persist scope, decisions,
source pointers, checks, next work and unresolved findings in Beads/ledger.
Afterward re-read the ledger and governing phase, query Beads, and inspect Git.
A failed phase gate, failed exit criterion or material unresolved blocker stops
advancement. Diagnose authorized in-scope failures; do not keep dispatching without
new evidence or bypass the task engine's budget and safety boundaries.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 468k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |