Menu
Skills in this repository
SkillsMP has collected 178 skills from DylanCkawalec/opgrok. Open a skill to review its source and details.
DylanCkawalec/opgrokShowing 40 of 178 collected skills.
Routes work to the correct SuperGrok role inside the debug domain (root-cause analysis, minimal fixes). Activates when selecting a debug specialist, browsing debug agents, or invoking /cat-debug. Differentiator: category index only — does not perform debug…
Isolates production failures to a single root site via repro-first RCA, then lands the smallest correct fix unit. Activates on crash stacks, flaky tests, CI-only failures, log anomalies, or /debug-smith. Differentiator: refuses any patch until a deterministic…
Traces production and test failures from symptom through hard evidence to a single root cause and minimal fix. Activates on crash stacks, flaky tests, CI-only failures, or /debug-trace. Differentiator: blocks any patch until a one-variable repro flips; causal…
Audits CI/CD workflows, container images, deploy pipelines, and runbooks against an explicit pass/fail checklist with path:line evidence. Use when fixing GitHub Actions failures, hardening Dockerfiles, or reviewing rollback readiness; invoke via…
Builds and repairs CI/CD, containers, deploy pipelines, runbooks, and platform config by forging the full commit→build→test→deploy path before hardening edges. Activates on failing Actions/GitLab jobs, Dockerfiles, Helm/Kustomize deploys, rollback runbooks,…
Maps CI/CD graphs, container bases, deploy targets, and secret surfaces before any pipeline or infra edit. Activates on failing Actions/GitLab jobs, Dockerfile/compose reviews, runbook gaps, or /devops-scout. Differentiator: traces workflow DAGs and image…
Finalizes DevOps delivery artifacts: gates CI green, freezes image digests and workflow pins, attaches rollback runbook, marks handoff-ready. Use when sealing GitHub Actions/Docker/K8s changes, fixing flaky jobs, or on /devops-seal. Differentiator: refuses…
Routes work to the correct SuperGrok role inside the devops domain (CI/CD, containers, deploy). Activates when selecting a devops specialist, browsing devops agents, or invoking /cat-devops. Differentiator: category index only — does not perform devops…
Builds and repairs the smallest correct CI/CD unit, container stage, or deploy automation that meets the brief. Activates on failing GitHub Actions/GitLab CI, Dockerfile hardening, compose/K8s manifests, runbooks with rollback, or /devops-smith.…
Traces CI/CD and platform failures via RCA: symptom log → evidence artifact → root config/code → fix with rollback. Activates on failing GitHub Actions/GitLab jobs, flaky matrix builds, broken Dockerfiles, deploy runbooks, or /devops-trace. Differentiator:…
Audits README, API docs, runbooks, and operator guides by checklist: every command, path, flag, and env var is cross-checked against repo source and scored pass/fail with path:line evidence. Activates on /docs-audit or requests to verify quickstarts, API…
Forges README, API reference, runbooks, and operator guides by building the verified end-to-end path first, then hardening edges. Activates on /docs-forge or docs work such as rewrite quickstart with repo-checked commands, document an API route with live…
Maps README, API docs, runbooks, and operator guides against live repo truth before any edit: inventories entrypoints, verifies every command/path flag against source, and flags drift hotspots. Activates on /docs-scout or briefs like "Rewrite README…
Finalizes README, API docs, runbooks, and operator guides by sealing only after every command, path, flag, and env var is proven against the live tree. Activates on /docs-seal or briefs like "Rewrite README quickstart with verified commands", "Document API…
Routes work to the correct SuperGrok role inside the docs domain (README, API docs, runbooks). Activates when selecting a docs specialist, browsing docs agents, or invoking /cat-docs. Differentiator: category index only — does not perform docs implementation…
Builds the smallest verified doc unit (README quickstart, API route page, runbook, operator guide) by cross-checking every command, flag, path, and env var against live source and --help output. Activates on /docs-smith or briefs like "rewrite README with…
Traces documentation drift via RCA: symptom → evidence → root → fix, verifying every command, path, flag, and env var against the live repo before rewrite. Activates on README/API/runbook/operator-guide tasks, stale quickstarts, or /docs-trace.…
Audits eval artifacts via frozen checklists: rubrics, judge contracts, harness journals, and pass gates. Activates on /eval-audit or requests to score a run, build a multi-dimension rubric, or verify threshold integrity. Differentiator: predeclared-threshold…
Designs and runs evaluation systems—rubrics, judges, harnesses, pass gates, scoring— via the forge method: wire the full e2e path (dimensions → threshold → score → gate) before hardening edge cases. Activates on rubric/judge/harness briefs or /eval-forge.…
Maps evaluation structure before any score is assigned: freezes dimensions, scales, and pass thresholds, then inventories evidence sources for rubrics, judges, harnesses, and gates. Activates on rubric design, judge-node contracts, harness pass-gate layout,…
Finalizes evaluation runs by locking predeclared thresholds, freezing rubric versions, and emitting irreversible PASS/FAIL with per-dimension evidence. Use when sealing harness scores, judge contracts, or rubric handoffs, or on /eval-seal. Differentiator:…
Routes work to the correct SuperGrok role inside the eval domain (rubrics, judges, harnesses). Activates when selecting a eval specialist, browsing eval agents, or invoking /cat-eval. Differentiator: category index only — does not perform eval implementation…
Builds minimal eval units—rubrics, judge contracts, harness pass-gates, score sheets—with predeclared thresholds and per-dimension evidence rules. Activates on /eval-smith or briefs like “4-dimension rubric for harness run,” “freeze judge node contract,”…
Builds symptom→evidence→root→fix causal chains for rubrics, judges, harnesses, and pass gates. Activates on /eval-trace or tasks like scoring a harness run against a frozen threshold, designing a judge-node contract, or RCA of dimension drift. Differentiator:…
Authors falsifiable Winning Conditions and refusal gates for OPGROK harness packages and SuperGrok catalog law. Activates on /leslie, harness sealing, catalog validation, WC authorship, or specification gates. Differentiator: specifies admissible behaviors…
Audits algorithms, numerics, complexity, and formal claims via explicit checklist scoring (PASS/FAIL per item with evidence). Use when deriving complexity, verifying formulas on boundaries, or implementing algorithms with test vectors; triggers on…
Derives algorithms, numerics, complexity bounds, and formal arguments by forging the full problem→model→verify path before edge hardening. Activates on concrete math asks (time/space complexity of a named algorithm, formula verification on boundaries,…
Maps algorithm structure, recurrence relations, and numeric domains before any edit or proof claim. Activates on complexity derivation, boundary-vector design, formula verification, or /math-scout. Differentiator: rejects hand-waved Big-O and unstated…
Finalizes math and algorithm work by locking definitions, attaching proof sketches or executable test vectors, and emitting a hard WIN gate. Use when complexity bounds, numeric results, or formal claims must be frozen for handoff. Triggers on /math-seal and…
Routes work to the correct SuperGrok role inside the math domain (algorithms, numerics, formal reasoning). Activates when selecting a math specialist, browsing math agents, or invoking /cat-math. Differentiator: category index only — does not perform math…
Builds the smallest verified math/algorithm unit: definitions, lemma or function, then test vectors or proof sketch. Activates on Derive time complexity of a concrete algorithm, Implement an algorithm with boundary checks, Verify a formula on edge inputs, or…
Traces algorithmic and formal failures via RCA: symptom → evidence → algebraic/numeric root → fix. Activates on Derive time complexity of a concrete algorithm, boundary-formula checks, induction gaps, or /math-trace. Differentiator: forces definition-first…
Audits MCP server wiring, tool schemas, and authenticated calls against an explicit pass/fail checklist. Activates when discovering or invoking Linear/GitHub/custom MCP tools, diagnosing schema/auth gaps, or when the user invokes /mcp-audit. Differentiator:…
Wires MCP servers end-to-end: schema discovery, auth-bound tool calls, multi-tool data contracts. Use when integrating Linear/GitHub/custom MCP tools, diagnosing schema/auth gaps, or invoking /mcp-forge. Differentiator: search_tool-before-use_tool discipline…
Maps MCP server inventories, tool JSON-schemas, and auth surfaces before any use_tool call. Activates on Discover/call Linear or GitHub MCP tools, schema-gap reports, new server config wiring, or /mcp-scout. Differentiator: schema-first recon that never…
Finalizes MCP tool sessions: freezes verified search_tool/use_tool transcripts, locks schema-matched inputs, and emits handoff-ready evidence when Linear/GitHub/custom MCP calls succeed or auth/schema gaps are explicit. Activates on /mcp-seal or when sealing…
Routes work to the correct SuperGrok role inside the mcp domain (MCP servers, schemas, discovery). Activates when selecting a mcp specialist, browsing mcp agents, or invoking /cat-mcp. Differentiator: category index only — does not perform mcp implementation…
Builds the smallest correct MCP tool-use unit: discover schema, bind exact inputSchema params, execute one authenticated call, surface auth/schema gaps without fabrication. Activates on Linear/GitHub/custom MCP tool wiring, missing-field reports, server…
Traces MCP tool failures via schema-first RCA: symptom → list_tools/search_tool evidence → root (missing field, auth scope, stale server) → fix. Activates on broken Linear/GitHub/Slack MCP calls, schema gaps, or /mcp-trace. Differentiator: never invents…
Owns the OPGROK Grok/xAI visual system under assets/: design tokens, deterministic SVG generation, and Grok Imagine atmospheric PNGs. Use for brand kits, UI icons, GitHub banners, empty states, protocol art, or /meta-asset-creator. Exact logos and chrome stay…