RonMizrahi
- 5 skills
- 0 followers
- 7 hours ago last updated
- ▌ View Run · ronmizrahi bundleGenerate a visual, LangSmith-style HTML view of an sdlc-graph run from its state file (docs/graph-runs/<run-id>/state.json) — the graph with the current node highlighted, next-step guards, the transition trace, milestones, retry counters vs bounds, and the skipped-gate ledger. Produces a self-contained snapshot HTML (openable and shareable anywhere). One invocation, no arguments: it works on the current directory, and asks which subproject when more than one has runs. Use this skill WHENEVER the user wants to SEE a graph run — "show me the run", "where is the graph now", "visualize the state", "open the run viewer", "what's the current state of the sdlc run", "generate the run report page" — or asks to watch a run live.
- ▌ Graph Run Reviewer · ronmizrahi bundleReview a finished or stuck sdlc-graph run — read its state file, its milestone journals and (when it was traced) every state it passed through, find what actually went wrong, and write the eval that would have caught it. Runs the offline auditor first for the mechanical defects, then looks for the classes no static check reaches: an observation that contradicts the journal, a gate that passed too cleanly, a ledger entry that does not match the tools present, a transition that was legal but wrong. Every finding is triaged as a spec defect, a run defect, or an eval gap — and every eval gap is closed with a real test, at the tier that catches it. Read-only with respect to the run: it never fixes, resumes, or edits the run it is reviewing. Invoked by a human, never automatically.
- ▌ Sdlc Graph Engineering Install · ronmizrahi bundleInstall graph engineering into a project: take an existing process — skills, a runbook, a README, a CI pipeline, or something the user can only describe out loud — and scaffold it into a working guarded graph with typed nodes, total exit guards, bounded retry loops, a durable run-state file, and a ledger recording every step that could not run. Writes the orchestrator skill and its model files into the target project, then proves the result sound by dry-tracing it before handing it over. Use this WHENEVER someone wants a multi-step process made resumable, auditable or reliable — "set up graph engineering here", "turn my workflow into a graph", "make this process a state machine", "my agent loses track halfway", "it retries forever", "it says it did things it didn't do", or when a prose runbook has grown too many conditionals to follow. Builds from scratch for any domain and needs no particular skills installed.
- ▌ Onboarding · ronmizrahiProduce the sdlc-graph tooling checklist — walk the plugin's dependency roster, report tool by tool what this session can actually invoke and what it cannot, print the install line for each gap, and say what a run does without it. Every tool on that roster is third-party and optional: the graph requires git and nothing else, and every absence is recorded in `skipped_gates[]` while the run continues. Invoked at the start of every sdlc-graph run, and by a human any time the question comes up. Purely advisory — it installs nothing, gates nothing, and never stops or delays a run.
- ▌ Sdlc Graph · ronmizrahi bundleDrive a feature end to end through the full SDLC as an explicit guarded graph — spec, plan, branching strategy, then per milestone: implement, test, the two quality gates, PR, CI, merge — then close-out and an independent QA pass. Keeps typed run state in `docs/graph-runs/<run-id>/state.json` so a run survives context loss and resumes exactly where it stopped, enforces a bound on every retry loop, and records every gate that could not run instead of silently passing it. Invoke explicitly to start a run, resume one ("where were we?", "next milestone", "continue the run"), or report its status. This skill is never auto-triggered — running an entire SDLC is not something to enter on a description match.