rishbjain1

goal

Lock in an end-state instead of a step. Plan, execute, self-verify, and debug autonomously until every success criterion is met. The user sets one outcome and walks away — no babysitting.

Turn any linear workflow into a graph and run it. Walks a workflow you already do step-by-step, applies the fake-edge test to find the waits that carry no data, audits for hidden edges (shared files, browser, rate-limited APIs), picks the shape (chain / diamond / router / cycle), emits a GRAPH SPEC, then runs it as a coordinated fleet via the Workflow tool with caps and fresh-context verifiers. Domain-agnostic — works on research, discovery pipelines, content pipelines, repo sweeps, analytics, launches, not just code. Triggers - "graph loop", "/graph-loop", "draw the graph", "graph this workflow", "fake edge test", "can this run in parallel", "make this parallel", "fan this out".

Code & Development Listed

karpathy-guidelines

Field notes for getting a language model to write code you will not rewrite. Reduces the predictable LLM coding mistakes — writing before reading, silent wrong assumptions, overcomplication, scope creep, unverified work, papering over bugs. Use when writing, refactoring, debugging, or reviewing code. Ten earned rules; the throughline is that the model is fast at generating plausible code and slow to notice plausible is not correct, so the discipline comes from the process around it. Source — Andrej Karpathy, CLAUDE.md field notes. Activates on "karpathy", "clarity", "karpathy mode", "be careful", "think first", "/karpathy", or any coding task that warrants discipline.

outer-loop

Bilevel loop engineering — put a loop on top of a loop. When a recurring process already runs (a sweep, a pipeline, a review cycle) but keeps getting stuck the same way, this reads its traces, finds where the SEARCH is stuck rather than where the output is wrong, and proposes a change to the mechanism itself. Based on Karpathy's AutoResearch loop plus Qu & Lu's "Bilevel Autoresearch" (arXiv 2603.23420), which got a verified 5x over a single loop using the SAME model at both levels — the gain is architectural, not intelligence. Requires an inner loop with traces and a metric; refuses to run without them. Triggers - "outer loop", "/outer-loop", "bilevel", "loop on the loop", "why does this loop keep getting stuck", "improve the process not the output", "meta-loop".

fable

Run ONE hard reasoning/planning task on Fable 5 as a subagent, without leaving your current (Opus/Sonnet) seat. Use when you're on a cheaper model mid-session and hit a problem that genuinely needs frontier judgment — a subtle design call, a deep analysis, a strategic tradeoff — and want Fable's depth on just that piece. The main loop stays on the cheap model; only this one sub-task runs at Fable rates; only its distilled answer returns to context. Triggers - "fable this", "reason this on fable", "run this on fable", "/fable", "fable-reason".

improve

Survey any codebase as a senior advisor and produce prioritized, self-contained implementation plans for OTHER models/agents to execute. Strictly read-only on source code — never implements, fixes, or refactors anything itself. Use when asked to audit a codebase, find improvement opportunities (bugs, security, performance, test coverage, tech debt, migrations, DX), suggest features or where to take the project next (roadmap, product direction), or generate handoff plans for another agent to implement.

conversion-system

Turn recurring manual work into trusted automation via 5 steps: capture → qualify → spec → build → trust ladder. Use when the user says /conversion-system, 'should we automate this', 'capture this task', 'add to automation registry', or when any task is noticed being done manually for the 2nd+ time. Also the gate every NEW automation must pass before being built.

handoff

Dump full session state to a persistent memory file in the active project so context can be cleared and any future Claude session restores instantly. Use when user says "handoff", "save context", "dump memory", "clear safely", "remember everything", or before /clear.

manager

Manager mode — hybrid orchestration for ANY work (coding, brainstorming, reasoning, research, drafting). Routes each sub-task to the cheapest model that meets the standard and reserves Fable 5 for a short frontier-only list, so leaving it on standing Fable barely touches Fable tokens. Codex (GPT 5.6 sol high) takes heavy coding and heavy reasoning; sonnet takes brainstorm / draft / medium reasoning / light code; haiku takes mechanical reading. Trigger: /manager, "manager mode", "orchestrate with the best model", "mixed fleet", or any multi-part task — not only coding.