docs: refresh verification skills and add allowlist workflow

[tamtamchik]

Repository skills contained stale config requirements, references to removed verifier functions, and guidance that could hide unverified bytecode. Update the four existing workflows against the current implementation and add a dedicated allowed-diffs workflow for justified exceptions and wildcard tightening.

Keep detailed explorer payload mappings and diagnostic recipes in linked references, including an executable immutable-rule coverage check. Preserve manual invocation settings and argument hints, and update the workflow index in CLAUDE.md.

Clarify the limits of source hunk rules, the prerequisites for constructor overrides inside allowlist rules, and the deployment-simulation exception accepted by bytecode wildcards. Keep config review separate from edits and live diagnostics, and correct the architecture description of immutable analysis. Explain recursive filename lookup, missing-source suggestions, token fallback and the simulation gas setting.

Validation

• Five skills: metadata and local reference links validated.
• Immutable-rule example executed: expected difference accepted and five unrelated changes rejected.
• 69 config, allowlist, explorer and HTTP tests passed.
• Independent reviews checked runtime accuracy, workflow completeness and removed guidance; reproduced that source rules match coordinates without pinning text.
• A further read-only Claude audit checked all five skills against runtime code and tests; confirmed findings were corrected after inspecting the implementation.
• Six read-only scenarios exercised through Claude CLI; follow-up runs checked CREATE2 reasoning, metadata evidence, hostname matching, config review and report fields. These checks do not establish live-chain support or reliability across models.

[Copilot]

🔵 Needs a closer look

Readiness assessments conflict, and the validation guidance has an unresolved behavioral inaccuracy.

Pull request overview

Refreshes verification workflows and adds guidance for tightly scoped allowed differences.

Changes:

• Updates config, debugging, deployment, and explorer workflows.
• Adds allowlist guidance and diagnostic references.
• Refreshes workflow and tooling documentation.

File summaries

File	Description
CLAUDE.md	Updates workflow and architecture references.
.claude/skills/validate-config/SKILL.md	Modernizes validation guidance. Nit: clarify that any: true can allow caught simulation errors.
.claude/skills/new-config/SKILL.md	Revises deployment-config creation guidance.
.claude/skills/debug-diff/SKILL.md	Aligns debugging with JSON reports.
.claude/skills/debug-diff/references/diagnostic-recipes.md	Adds focused diagnostic recipes.
.claude/skills/allowed-diffs/SKILL.md	Adds narrowly scoped exception guidance.
.claude/skills/allowed-diffs/references/check-immutable-rule.md	Adds an immutable-rule coverage check.
.claude/skills/add-explorer/SKILL.md	Updates explorer integration guidance.
.claude/skills/add-explorer/references/response-mapping.md	Documents payload normalization.

Review details

Suppressed comments (1)

.claude/skills/validate-config/SKILL.md:35

• This overstates the failure behavior: process_config marks a caught DeploymentSimulationError as allowed when the contract has a bytecode any: true rule. Calling all caught bytecode errors failed can make config reviews miss that this wildcard suppresses simulation failures; document the exception explicitly.

`load_config` validates `allowed_diffs` through `diffyscan/utils/allowed_diffs.py`. Schema validity does not justify a rule: inspect reason and scope. Use [allowed-diffs](../allowed-diffs/SKILL.md) when tightening or adding exceptions. `fail_on_bytecode_comparison_error: false` can let an outer contract error continue, but caught bytecode errors still produce failed results.

• Files reviewed: 9/9 changed files
• Comments generated: 0
• Review effort level: Balanced

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[Copilot]

🟢 Approval recommended

The reviewed documentation changes have no unresolved approval-blocking issues.

Review details

• Files reviewed: 9/9 changed files
• Comments generated: 0 new
• Review effort level: Balanced

[TheDZhon]

No actionable findings in PR #216 (#216), reviewed at 2d01b85. Rechecked GitHub; the head is unchanged.

The revised workflows match the runtime behavior, including allowlist restrictions, constructor overrides, and simulation-error handling.

Validation:

• 410 tests passed
• Five skill metadata records and 32 local links/anchors validated
• Embedded immutable-rule example passed all six assertions