conch-review-smoke

name	conch-review-smoke
description	Run minimal runtime validation for Conch and separate environment-readiness problems from product behavior problems. Use when the user wants to actually bring Conch up, run a smoke path, or verify whether failures are environmental or implementation-related.

Conch Review Smoke

Use this skill when the user wants runtime evidence rather than a purely static review.

Inputs

• target_repo: path to the Conch repository
• optional mode: probe-only or execute

Default target repo: /home/stopire/Conch Default mode: probe-only

Load These Assets

• policies/severity.md
• policies/report-template.md
• reviewers/sandbox-e2e-validator.md

Support reviewers when interpretation needs them:

• reviewers/lifecycle-reviewer.md
• reviewers/docs-implementation-auditor.md

Use these scripts:

• scripts/discover_conch.py
• scripts/run_smoke.py
• scripts/normalize_findings.py when you need a unified findings envelope

Workflow

1. Discover the repo shape

Run:

python3 scripts/discover_conch.py <target_repo>

2. Run readiness probe

Run:

python3 scripts/run_smoke.py <target_repo> --mode probe-only

If readiness fails, stop there and report the result as an environment-readiness outcome unless there is clear evidence of a product defect.

3. Run execution mode when requested

If the user asked for execution mode and readiness passed, run:

python3 scripts/run_smoke.py <target_repo> --mode execute

4. Interpret the result

Use sandbox-e2e-validator as the primary reviewer for diagnosis. Pull in:

• lifecycle-reviewer when a runtime failure suggests a lifecycle contract problem
• docs-implementation-auditor when the runtime failure contradicts operator-facing instructions or examples

5. Report a step-by-step smoke outcome

Use policies/report-template.md as the output skeleton.

Produce:

• environment readiness status
• smoke steps attempted
• pass/fail/skipped per step
• environment versus product diagnosis
• current smoke-runner limitations
• recommended next action

Constraints

• Be explicit about what was actually executed versus only probed.
• Do not overclaim product correctness from a shallow smoke pass.
• If deeper static reasoning is required, recommend conch-review-deep-review.