AgentStack
Browse Sign in
Browse Why AgentStack Sell Docs
Sign in
SKILL verified MIT Self-run

Claudex Grill With Docs

skill-tschuehly-skill-incubator-claudex-grill-with-docs · by tschuehly

Run a documented grilling session in claudex with Claude leading the interview and the GPT/Codex-family grill-advisor challenging the opening design tree and auditing the final decisions. Use when the user wants Claude and Codex to stress-test a plan or design together while recording ADRs and domain terminology.

No reviews yet
0 installs
29 views
0.0% view→install

Install

$ agentstack add skill-tschuehly-skill-incubator-claudex-grill-with-docs

✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

View the full security report →

Verified badge

Passed review? Show it. Paste this badge into your README, it links to the public security report.

AgentStack Verified badge Links to your public security report.
[![AgentStack Verified](https://agentstack.voostack.com/badges/verified.svg)](https://agentstack.voostack.com/security/report/skill-tschuehly-skill-incubator-claudex-grill-with-docs)

Reliability & compatibility

Security review passed
0 installs to date
no reviews yet
2mo ago

Declared compatibility

Claude CodeClaude Desktop

Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.

Preview Execution monitoring

We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.

How agent discovery & health will work →
Are you the author of Claudex Grill With Docs? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

Claudex Grill with Docs

Compose /grilling, /domain-modeling, /atelier, and /agent-orchestration. Claude owns the conversation and documentation; the grill-advisor agent (gpt-5.6-sol) acts as the Codex advisor. This skill fixes that advisor binding; use /model-orchestration only for broader routing and quota policy outside this composition.

Workflow

  1. Confirm the session is running under claudex and the grill-advisor agent is

available. If it is unavailable or its model fails to resolve, explain that the cross-family session cannot run and do not claim that Codex participated.

  1. Establish the target. If the user named a doc, branch, handoff file, or topic,

use it — a handoff passed as an argument is the target; read it first. If none is given, ask what to grill. Do not spelunk the repo to guess the topic.

  1. Gather facts from the codebase and supplied documents. Distill the plan,

constraints, known facts, and unresolved decisions; do not give the advisor raw file dumps.

  1. Before asking the first question, start one persistent grill-advisor. Ask it to

identify the consequential assumptions, missing branches, contradictions, and highest-value questions, with a recommended answer for each.

  1. Use those findings as input to a /grilling session that maintains docs via

/domain-modeling. Follow both contracts: work one decision at a time, recommend an answer, resolve discoverable facts without asking the user, and maintain the ADRs and glossary as decisions settle. Present each decision richly — the tension, a worked example, the verified facts (with file:line), and your recommendation — never a bare one-line question. Present the decisions on an /atelier surface with one Review Unit per decision. Apply /agent-orchestration for collaborator reuse, compact handoffs, durable state, continuity, and one human-facing voice.

  1. Synthesize the advisor's advice into the grill; do not invoke the advisor before

every question.

  1. Before declaring shared understanding, distill the settled decisions, rationale,

glossary, and remaining uncertainties. Continue the same grill-advisor with that compact delta for an adversarial completeness check.

  1. If the final check finds a material gap, resume the grill one decision at a time.

Otherwise, present the intended-state documents for confirmation.

  1. Do not implement the plan, or ratify ADRs and glossary terms as decided, until

the user explicitly confirms shared understanding.

Treat the two advisor calls as the default budget. Add another only when new information materially invalidates an earlier branch.

AFK / autonomous convergence

When the user says they are away and asks you to iterate to shared understanding without them, the advisor stands in for the human interviewer:

  • Run grill-advisor rounds (distill → challenge → correct → re-check) until the

advisor confirms shared understanding or you reach the round budget. Apply /agent-orchestration to continue the advisor and preserve resumable state.

  • Leave every decision as an open proposal on the surface; mark nothing decided.

ADR and glossary ratification still waits for explicit human signoff (step 9).

  • End with the durable grill record ready for resumption.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet, be the first.

Versions

  • v0.1.0 Imported from the upstream source.