ronniepinnell
- 58 skills
- 0 followers
- 8 hours ago last updated
- ▌ Door · ronniepinnellOne-way-door decision triage. Sorts any decision into reversible (just pick) vs irreversible (slow down, enumerate lock-in). Use before schema changes, ID formats, API contracts, auth models, naming, tech-stack picks, or whenever a choice feels weighty.
- ▌ Gate · ronniepinnellNumeric kill-gate authoring. No plan, experiment, or migration is accepted without a measurable abort condition — what's measured, what threshold, what happens on fail. Use at plan time, before long-running work, and when adjudicating "is this good enough to continue".
- ▌ Drift · ronniepinnellSpec vs code-as-built diff. Compares what a spec/doc/contract declares against what the code and live system actually do, and reports which one is lying. Use when behavior surprises you, before building on a spec, or as a periodic honesty sweep.
- ▌ Sweep · ronniepinnellMassive multi-agent audit sweep — the generalized FAB-project pattern. Fans out parallel domain auditors over a whole system (codebase, data stack, spec tree, product surface), adversarially verifies findings, and converges to a graded report where every finding lands as a ticket, gate, or patch. Use for "audit everything", pre-launch reviews, post-chaos reconciliation, or grading a stack against best-in-class.
- ▌ Think · ronniepinnellStructured thinking moves for when you're stuck, the answer feels too easy, or a problem needs depth — invert, second-order, base-rate, analogy, constraint-flip, decompose. Routes a situation to the right move and forces it to completion. Use for design deadlock, suspicious consensus, novel problems, or "we've always done it this way".
- ▌ Refute · ronniepinnellAdversarial claim verification. Before accepting any "it works / it's done / it's fixed" claim, construct the concrete input that would break it and run it. Use before commits, PR creation, closing issues, or whenever a completion claim is made.
- ▌ Verdict · ronniepinnellAppend-only judgment ledger. Every DOOR / GATE / PREMORTEM / REFUTE / DRIFT verdict line gets logged to one grep-able file so "show me every gate we overrode" is a query, not archaeology. Use whenever a judgment skill produces a verdict, and to query past verdicts.
- ▌ Altitude · ronniepinnellRight-layer check before coding a fix. Locates which layer a problem actually lives at (data model, storage, API, business logic, UI, process/people) so the fix lands at the cause's layer, not the symptom's. Use before fixing any bug, and whenever the same bug keeps coming back.
- ▌ Backfill · ronniepinnellRetroactively grade merged PRs against the claim-evidence discipline — which done-claims shipped with evidence, which shipped unproven. Zero-LLM (gh CLI); optionally seeds the verdicts ledger with UNVERIFIED BACKFILL rows. Use when adopting the toolkit on a repo with history, before trusting old "done"s, or to measure how the discipline is trending.
- ▌ Escalate · ronniepinnellJudgment escalation queue. When a session hits a judgment-dense question it shouldn't wing — a one-way door with no precedent, a borderline gate, a spec-vs-code conflict, a calibration miss — it stops that thread, queues the question to .claude/escalation-queue.md with a provisional answer, and ships the rest. Use /escalate burn to batch-adjudicate the queue with a frontier model or human.
- ▌ Fix Pool · ronniepinnellFindings→fixes pooler for a coordinator session. Polls the task manager for new issues (walkthrough/tester-filed or manual), triages each (auto-fix / investigate / hold-for-operator / duplicate), and dispatches throttled background fix agents that stop at PR. Use when a build-hub session runs alongside human testing slots and new findings should flow into the builder workflow without the operator relaying them. Arm with `/fix-pool arm`, check `/fix-pool status`, stop with `/fix-pool drain`.
- ▌ Judgment · ronniepinnell bundleIndex and installer for the judgment toolkit — six procedure skills (/refute /door /gate /drift /altitude /premortem) and three mechanical hooks (claim-evidence, spec-citation, scope-creep) that bank frontier-model judgment as forced procedure any model can run. Use to see the toolkit, pick the right tool, or install the hooks into a project.
- ▌ Calibrate · ronniepinnellScore how past confidence aged. Samples old tagged claims ([HIGH]/[MED]/[LOW], gate predictions, premortem risks, estimates) and checks them against what actually happened. Use monthly, after a milestone closes, or whenever confidence tags start feeling like decoration.
- ▌ Precedent · ronniepinnellPrecedent lookup before deciding. Greps prior rulings — the verdicts ledger, decision stores, the adjudicated escalation queue — for anything bearing on a pending decision, then either FOLLOWS the precedent or explicitly DISTINGUISHES the case. Use before any /door call, gate override, or decision in a domain the project has ruled on before. No silent departures.
- ▌ Premortem · ronniepinnellWrite the incident report before shipping. Assume the work shipped and failed 3 months from now; write the post-mortem first, then harden the top risks. Use before locking a design, launching a feature, running a migration, or dispatching autonomous work.
- ▌ Merge Train · ronniepinnellMerge every open PR that is mechanically safe — CI green, approved, mergeable, autonomy label present, no unacknowledged override markers — in base-first order; everything else is reported with the exact fact it is missing. Use at end of week, before a release, or whenever open PRs have piled up. Dry-run by default.
- ▌ Audit Skills · ronniepinnellEfficiency audit of a skill library on four measured axes — token cost, ceremony/duplication bloat, trigger-description precision, and overlap clusters — plus usage axes when invocation telemetry exists. Every claim carries a number; findings land as diffs and budgets, not advice. Use before tightening skills, after adding many, or as a periodic library health check.
- ▌ Brief · ronniepinnell> **MCP Tool Map (Gemini/Codex):** See `.claude/skills/_shared/mcp-tool-map.md` for tool name equivalents. Linear: `get_issue`/`update_issue`/`create_issue`/`list_issues`/`search_issues`. Supabase: use `python3 {scripts_dir}/infra/run_sql.py "<SQL>"` via Bash.
- ▌ Handoff · ronniepinnellContext-rot prevention: write a structured handoff document capturing the current session state so you can /clear and continue in a fresh session without losing context. Use when: context >70%, session >15 turns, hitting limits, or switching tasks mid-work. Invoke: /handoff [optional-filename]
- ▌
- ▌
- ▌ Intake · ronniepinnellImport external skills and agents into shared-config from a git URL or local path. Security-scans each unit, stamps origin/source provenance, dedupes name collisions, and updates the catalog. Use when asked to "intake", "import a skill", "add this agent", "pull in skills from <repo>", or "ingest these skills".
- ▌ Continue · ronniepinnellResume interrupted session. Loads context, companion findings, and git state.
- ▌ Langfuse · ronniepinnellInteract with Langfuse and access its documentation. Use when needing to (1) query or modify Langfuse data programmatically via the CLI — traces, prompts, datasets, scores, sessions, and any other API resource, (2) look up Langfuse documentation, concepts, integration guides, or SDK usage, or (3) understand how any Langfuse feature works. This skill covers CLI-based API access (via npx) and multiple documentation retrieval methods.
- ▌ Ccb · ronniepinnell> **Factory mode:** If `CLAUDE_AUTO` is set in the environment, skip these instructions and instead run `{scripts_dir}/skills/ccb.sh "$@"` via Bash — the shell script handles autonomous dispatch.
- ▌
- ▌ Iterate Pr · ronniepinnellIterate on a PR until CI passes. Use when you need to fix CI failures, address CodeRabbit feedback, or continuously push fixes until all checks are green. Automates the feedback-fix-push-wait cycle. Use with /loop for continuous monitoring.
- ▌
- ▌ Wave · ronniepinnell bundleAssess current project state and produce (and optionally dispatch) the next N-slot factory run — a wave of parallel `/epic start` prompts with model routing, Operator-gated items flagged, and the critical-path decision surfaced. The repeatable replacement for the manual "give me the next run order by slot with prompts" ask. Use when planning (or launching) the next batch of parallel factory work.
- ▌ Sync Skills · ronniepinnellPush or pull changes to the shared shared-config repo. Shows shared repo status AND flags project overlays that have diverged from the shared version.
- ▌ Issue · ronniepinnell> **Factory mode:** If `CLAUDE_AUTO` is set in the environment, skip these instructions and instead run `{scripts_dir}/skills/issue.sh "$@"` via Bash — the shell script handles autonomous dispatch. (`scripts_dir` from project-context; default `scripts`.)
- ▌ Design · ronniepinnellInteractive design session. Agent team debates architecture, captures decisions, logs ideas.
- ▌ Project Init · ronniepinnellBootstrap any repo with a full AI-ready scaffold. Runs as /project-init (full interview), /project-init --bootstrap (fast 5-question path), /project-init sync (drift check), /project-init update <section> (targeted refresh), or /project-init setup <service> (walk through setting up Supabase, Doppler, Vercel, GitHub App, Linear, Railway, etc.)
- ▌
- ▌ Export Public · ronniepinnellDrive the public export layer — the machinery that turns this private repo into the downstream public repos (casper, refute-action, the awesome-style skills collection, …). Use when asked to "export public skills", "publish my skills", "check public export drift", "sync the public repos", "make the free repo", or "update the public skills repo".
- ▌ Trigger Tasks · ronniepinnellBuild AI agents, workflows and durable background tasks with Trigger.dev. Use when creating tasks, triggering jobs, handling retries, scheduling cron jobs, or implementing queues and concurrency control in the factory.
- ▌
- ▌ Cleanup · ronniepinnellMass reconciliation when things are out of whack. Audit, triage, fix, verify.
- ▌ Dossier · ronniepinnellFrontier dossier session — run a repo's deepest available review with the strongest available model and bank ALL of it in one committed, append-only file. Spec-truth pass, graded deep review, executable fix rulings, a FULL build-to-spec plan, product/feature/tech ideation with gap analysis, and a premortem. Use when frontier-model time is scarce and its judgment must survive as a permanent artifact; rerun periodically and diff against the last section.
- ▌ Pr Review Loop · ronniepinnellPoll a PR for CodeRabbit review, apply fixes (up to 3 rounds), then merge. Bounded autonomous loop — give it a PR number and walk away.
- ▌ Trigger Agents · ronniepinnellAI agent patterns with Trigger.dev — orchestration, parallelization, routing, evaluator-optimizer, and human-in-the-loop. Use when building LLM-powered factory tasks that need parallel workers, approval gates, tool calling, or multi-step agent workflows.
- ▌ Pipeline · ronniepinnell> **Factory mode:** If `CLAUDE_AUTO` is set in the environment, skip these instructions and instead run `{scripts_dir}/skills/pipeline.sh "$@"` via Bash — the shell script handles autonomous dispatch.
- ▌ Sync AI Context · ronniepinnellDiff and re-sync all AI context files (GEMINI.md, Cursor rules, Copilot instructions, rules/BASE.md) against AGENTS.md. Run after editing AGENTS.md or changing critical rules to propagate updates to all derived files.
- ▌
- ▌ Find Bugs · ronniepinnellFind bugs, security vulnerabilities, and code quality issues in local branch changes. Use when asked to review changes, find bugs, security review, or audit code on the current branch. ALWAYS run before committing.
- ▌
- ▌ Spec Audit · ronniepinnellCode-vs-spec comparator. Reads the source first-hand and compares it against written specification documents, classifying every divergence as absent, partial, wrong, or extra — with file:line on both sides and an explicit call on which artifact should change. Project rules (CLAUDE.md) outrank specs. Use before PRs, after spec'd feature work, or when spec drift is suspected.
- ▌ Sred Work Summary · ronniepinnellGo back through the previous year of work and create a Notion doc that groups relevant links into projects that can then be documented as SRED projects.
- ▌ Cycle Check · ronniepinnellWeekly planning ritual. Reviews last week's progress, upcoming work, and flags blockers across the active milestone.
- ▌ Rules Audit · ronniepinnellProject-rules enforcer. Reviews recent changes strictly against the binding instructions in CLAUDE.md (and any project rule checklist), flagging every deviation with the exact rule cited and a concrete fix. Deliberately narrow — conformance only, not general code quality. Use after any code change and before commit.
- ▌
- ▌
- ▌ Plan Milestone · ronniepinnell bundleBreak a milestone goal into epics and tasks with dependency ordering, acceptance tests, and task manager ticket creation. Works with any task manager via the adapter layer.
- ▌ Sred Project Organizer · ronniepinnell bundleTake a list of projects and their related documentation, and organize them into the SRED format for submission.
- ▌ Completion Audit · ronniepinnellSession/project-scope reality audit. Independently establishes how much of the claimed-done work is genuinely functional, cross-checked against every planning source, and produces a prioritized remediation plan. Use before committing or closing issues, when statuses say done but the system misbehaves, or whenever an honest project snapshot is needed. Supports --task <id> for single-task scope.
- ▌ Pragmatism Audit · ronniepinnellOver-engineering review. Examines recently written code for complexity that the project's actual scale and needs don't justify, and proposes the smallest design that still works. Use after implementing a feature or making an architectural decision, before completion review.
- ▌
- ▌ Validate Completion · ronniepinnellSingle-task completion verifier. When an implementer claims a task or feature is finished, establish whether the goal was genuinely achieved — by executing it when possible, by rigorous inspection when not. Compiling, existing, or green tests are not proof. Use immediately on any "done" claim, before the status is recorded.