Claude Skill

verification-methodology

Verify work against explicit criteria using direct, source-faithful evidence, reproducible checks, and clear verdicts. Use before declaring an artifact, implementation, or claim complete; do not use for exploratory research without pass/fail criteria.

LLM Mart · 0 points · 6 views 0 listing impressions 0 install-command copies
Virus-scanned Reviewed automatically before listing.

Full trust report

Download magnus919-agent-skills-verification-methodology-1a7d575.zip · 8 KB
magnus919/agent-skills 96 9 forks MIT Updated 2d ago
Part of magnus919/agent-skills — 145 skills

Install

skills CLI npx skills add https://github.com/magnus919/agent-skills/tree/main/verification-methodology
Claude Code claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install magnus919-agent-skills@llmmart
Git git clone https://github.com/magnus919/agent-skills.git

The skills CLI installs just this skill, for any of its supported agents. Claude Code installs the whole magnus919/agent-skills collection as a plugin from our marketplace. Git is the plain clone.

README

Verification Methodology

Replace completion claims with a disciplined evidence trail that shows what was checked, what passed, and what remains uncertain.

Why Install This Skill

Replace completion claims with a disciplined evidence trail that shows what was checked, what passed, and what remains uncertain. It preserves a practical method, local reference material, and reusable templates so an agent can do more than produce a generic answer.

Use it when the work needs a repeatable process and an inspectable result. It is portable across Agent Skills-compatible clients and does not require a profile system or a particular task orchestrator.

What You Get

Path What it provides
SKILL.md Trigger conditions, workflow, and guidance for loading deeper resources.
references/ Reference material: criteria-assessment.md, evidence-standards.md, verdict-template.md
evals/evals.json Regression cases for direct-source selection, bundled executable discovery, failure reporting, and explicit fallback handling.

Quick Start

Define the criteria, collect evidence from the requested source before any substitute, and produce a verdict using references/verdict-template.md.

Install or expose this directory using your agent's standard Agent Skills loading mechanism, then ask for work that matches the triggers below.

Triggers

  • Verify work against explicit criteria using direct, source-faithful evidence, reproducible checks, and clear verdicts. Use before declaring an artifact, implementation, or claim complete.
  • Requests involving the method, deliverables, or review process described in SKILL.md.
  • Work where a reusable template or reference from this skill would reduce avoidable mistakes.

Requirements

No runtime dependency.

Source and maintenance

This skill was extracted from magnus919/hermes-profiles at commit 867a555. The portable methodology was retained; Hermes-specific profile, orchestration, and memory assumptions were removed.

Skill manifest

Verification Methodology

Pass/fail assessment against pre-defined criteria.

The Verification Protocol

  1. Receive — restate the artifact, claim, or implementation being verified and the decision it will support.
  2. Assess criteria — convert requirements into observable pass/fail conditions; identify what would disprove each claim.
  3. Investigate — collect direct, reproducible evidence from the source named by the request and record commands, source locations, or source URLs. Own this collection when the source is accessible: do not ask the user to relay evidence you can retrieve yourself. Ask only for access you genuinely lack.
  4. Decide — mark each criterion passed, failed, blocked, or not applicable. Do not convert missing evidence into a pass.
    • When verification is split across lanes, no lane's PASS is the gate verdict. Wait for every required lane and reconcile conflicts before synthesis.
    • A later source-fidelity BLOCK supersedes an earlier prose, rendering, or checklist PASS for the same artifact hash. Immediately label earlier PASS packets as superseded so downstream workers cannot mistake them for the current verdict.
    • Pattern scans and internal consistency cannot substitute for checking factual transformations against the named primary source or source dossier.
  5. Report — use the verdict template to distinguish verified facts, assumptions, and remaining work.

Stop when every criterion has direct evidence or an explicit blocked/not-applicable verdict. Escalate when the criterion is ambiguous, evidence conflicts, or the required access is unavailable.

Source Fidelity

Treat the requested or configured local service as part of the verification criterion, not as an interchangeable topic label.

  1. Load the matching skill and use its documented executable or service path before adjacent integrations, generic web search, or public project sources.
  2. Use the path resolved by the skill itself. A missing global PATH entry does not prove that a bundled executable is unavailable.
  3. If the direct source fails, report the attempted command or endpoint and its exact failure. Do not silently substitute evidence from another source.
  4. Use a substitute only when the user requests broader context or explicitly accepts the fallback. Label substitute evidence as secondary and do not present it as the requested source's state.

Example: for “What’s new on Jellyfin?” in an environment with a configured Jellyfin skill and bundled CLI, query that server through the bundled CLI first. Home Assistant entities and public Jellyfin project activity answer different questions.

When not to use

Do not use this skill for open-ended exploration that has no artifact, claim, decision, or observable completion criterion. Use a research or discovery skill first, then return here when there is something falsifiable to verify.

Related Skills

  • playwright — browser-based E2E checks as a verification instrument: authoring and running specs against the real UI boundary.
  • documents — producing verification reports or evidence packets as PDF/Word/Excel/PowerPoint deliverables.

Reference Files

Reference When to load
references/criteria-assessment.md You need to evaluate whether work meets completion criteria
references/evidence-standards.md You need to judge whether evidence supports the claims made
references/magnus919-refine-to-ship-gate.md You are running the Magnus919 Refine-to-Ship verifier gate — 12 criteria, editorial change verification, output structure
references/verdict-template.md You need to produce a structured pass/fail/hold verdict

Magnus919 Refine-to-Ship Gate Criteria

12 criteria for the verifier profile. Each criterion maps to an observable, reproducible check.

All 12 Criteria

# Criterion How to Verify
1 Dash scan — zero em dash (U+2014), en dash (U+2013), horizontal bar (U+2015), or visible prose double-hyphen search_files for [\u2014\u2013\u2015] and \-\-. Double-hyphens in YAML frontmatter delimiters are OK.
2 Fact-check — all methodology claims map to source; no fabricated numbers, chronology, or universal claims Cross-reference article claims to source document sections. Search for \d+%, percent, average of, illustrative. Search for research proves, studies demonstrate.
3 Voice-check — Magnus fingerprint: conversational first-person, contractions, "But" pivots (not formal transitions), colons over semicolons, no consultant cadence Search for Furthermore, Moreover, Nevertheless, Consequently, Therefore, not only.*but also, triplet parallelism. Count colons vs semicolons (should skew heavily toward colons).
4 Oxford commas, spelling, grammar — American English, Oxford commas in series, no spelling errors Manual read of series. Check for consistent formatting.
5 No formulaic AI closing — zero "In conclusion", "To summarize", "In this article", generic motivational advice search_files for In conclusion, Ultimately,, To summarize, In this article, In this post.
6 Methodology-first — personal frame ≤ ~10% of article; rest is methodology Count paragraphs in frame vs body.
7 Human stake integrated — cognitive burden, expertise formation, transferred work, anti-surveillance, accountable authority Verify dedicated section or dispersed coverage of all dimensions.
8 Privacy/anonymization — zero company identifiers, role titles, named people, source filename, proprietary domain examples search_files for company name, product names, domain-specific terminology from source.
9 Frontmatter — title, slug, date, byline correct and value-identical to specification read_file lines 1–11.
10 Links resolve — each distinct URL appears once at first meaningful mention; all return 200 curl -s -o /dev/null -w "%{http_code}" each URL. Verify link text is at first meaningful mention.
11 Hugo build + routes — build exit 0; new route returns 200; old take-home-title route returns 404 hugo --quiet && echo EXIT:$?. curl both routes.
12 No duplicate source bundle — single directory, single index.md; no stale *take-home* directories ls the page bundle directory. find in content/posts for duplicate slug patterns.

Parent-Requested Editorial Changes

When the parent profile specifies editorial changes during gate recovery, verify each one is present before proceeding with the full criteria scan:

Change Type Verification Method
Fabricated illustrative numbers removed search_files for \d+%, percent, PRs? per, average of → zero hits
Tense correction search_files for the exact parent-specified phrase
Closing replacement search_files for the first and last sentence of the parent-specified closing

Verdict Rules

  • PASS: All 12 criteria met. Produce 00-index.md, 01-summary/verdict.md, 02-analysis/per-criteria-results.md.
  • BLOCK: Any criterion fails. Produce gap-details.md with specific fix instructions. See verifier-gate-recovery skill for remediation patterns.

Output Structure

/private/tmp/verifier-gate/<slug>-refine/
  00-index.md              — verdict, links to artifacts
  01-summary/verdict.md    — per-criterion pass/fail table
  02-analysis/
    per-criteria-results.md  — detailed evidence per criterion
    gap-details.md           — only if BLOCK, with remediation instructions

Portability

This skill is intentionally host-neutral. Use your agent's normal mechanisms to load the references, templates, and scripts listed here. Do not assume a particular profile system, task orchestrator, memory service, or response-handoff format.

Files (agent-skills)
  • evals
    • evals.json 4.4 KB 
      {
  "schema_version": 1,
  "skill_name": "verification-methodology",
  "evals": [
    {
      "id": "configured-jellyfin-server-first",
      "prompt": "The jellyfin skill is installed, its bundled CLI and valid server configuration are available, and the user asks: ‘What’s new on Jellyfin?’ Verify the answer.",
      "expected_output": "A source-faithful investigation that loads jellyfin and uses the skill-resolved bundled CLI to query recently added items on the configured server before considering adjacent integrations or public project sources.",
      "assertions": [
        "Loads the jellyfin skill before investigating service content.",
        "The first content query uses the bundled CLI's documented recent-items command through the path resolved by the skill.",
        "Does not query Home Assistant, GitHub, web search, or public Jellyfin project activity before attempting the configured server.",
        "Reports the configured server's results as server state rather than conflating them with project news."
      ]
    },
    {
      "id": "bundled-executable-not-on-path",
      "prompt": "The storage-audit skill resolves its executable to /opt/skills/storage-audit/scripts/storage-audit, but `which storage-audit` returns nothing. Verify the current storage report.",
      "expected_output": "An investigation that executes the skill-resolved absolute path instead of concluding that the tool is unavailable because it is absent from the global PATH.",
      "assertions": [
        "Attempts /opt/skills/storage-audit/scripts/storage-audit before trying a substitute.",
        "Does not treat a failed `which` lookup as proof that the bundled executable is unavailable.",
        "Records the exact executable path used as reproducible evidence.",
        "Bases the verdict on the bundled tool's result or exact failure."
      ]
    },
    {
      "id": "direct-source-failure-no-silent-substitution",
      "prompt": "Use the configured inventory CLI to verify which laptops are currently enrolled. The direct command exits with an authentication error, while a public vendor status page is reachable.",
      "expected_output": "A blocked or hold verdict that reports the exact direct-source authentication failure and identifies the access needed, without presenting the public vendor status page as evidence of local enrollment state.",
      "assertions": [
        "Reports the attempted inventory command or endpoint and the authentication failure.",
        "Marks local enrollment verification blocked or on hold rather than passed.",
        "Does not substitute the public vendor status page for the requested inventory source.",
        "States what access or configuration would resolve the block."
      ]
    },
    {
      "id": "explicit-broader-context-allows-secondary-source",
      "prompt": "First verify what version my configured Jellyfin server runs, then compare it with the latest upstream Jellyfin release notes.",
      "expected_output": "A two-source answer that queries the configured server first, then consults an authoritative upstream release source because the user explicitly requested the comparison, with each source and claim clearly labeled.",
      "assertions": [
        "Queries the configured Jellyfin server for its installed version before the upstream comparison.",
        "Uses an authoritative upstream release source for the latest release claim.",
        "Labels local server evidence and upstream project evidence separately.",
        "Does not imply that the latest upstream version is already installed locally."
      ]
    },
    {
      "id": "requested-trakt-source-before-adjacent-catalog",
      "prompt": "A configured Trakt skill documents a canonical CLI and credential source. Verify the user's Trakt watchlist. A movie-catalog integration is also available.",
      "expected_output": "A Trakt-first investigation that follows the skill's documented CLI and credential path, without switching to the adjacent movie catalog unless the user explicitly accepts that fallback.",
      "assertions": [
        "Loads the Trakt skill and follows its documented executable and credential discovery path.",
        "Attempts the Trakt watchlist query before any adjacent movie-catalog query.",
        "Does not replace a Trakt authentication failure with results from another catalog.",
        "Reports Trakt evidence or the exact Trakt failure as the basis for the verdict."
      ]
    }
  ]
}
  • references
    • criteria-assessment.md 1.2 KB 
      # Criteria Assessment

## The Verifier's Question

Not "is this good?" — that's a reviewer's question. The verifier asks: "does this meet the criteria?"

## Criteria Types

| Type | Example | Assessment |
|------|---------|------------|
| **Binary** | "All tests must pass" | Pass/Fail — no gray area |
| **Threshold** | "At least 80% test coverage" | Pass/Fail with a number |
| **Judgment** | "The argument must be coherent" | Requires explanation but still binary |
| **Edge case** | "Must handle null input" | Pass only if explicitly tested |

## Assessment Protocol

1. **Identify all criteria** from the original brief or quality gate
2. **Rate each criterion** independently — don't let one failure influence others
3. **Distinguish failure from uncertainty.** If you can't verify a criterion because information is missing, flag it as uncertain, not failed
4. **Report partial compliance.** If a criterion is partially met, state what's met and what's not

## Verdict Decision

- **PASS:** All criteria met. No exceptions.
- **HOLD:** Most criteria met but some require clarification or information is missing.
- **FAIL:** One or more criteria clearly not met, or evidence contradicts claims.
    • evidence-standards.md 1 KB 
      # Evidence Standards

## Evidence Levels

| Level | Standard | Example |
|-------|----------|---------|
| **Primary** | Direct observation, original data, reproducible experiment | Test output showing a passing result |
| **Secondary** | Report from a credible source, cited with provenance | Quoting an API documentation reference |
| **Circumstantial** | Indirect evidence that suggests but doesn't prove | Logs showing the system was up, but not that it processed correctly |
| **Assertion** | Claim without supporting evidence | "The system should work" |

## Verification Rules

- Claims require evidence at the level specified by the criteria
- A primary-source criterion cannot be satisfied with secondary evidence
- Missing evidence for a criterion is a HOLD, not a FAIL
- Contradictory evidence (claim says X, evidence shows Y) is a FAIL

## Uncertainty Handling

When evidence is insufficient:
1. Document what evidence is available
2. Document what evidence would be needed
3. Report as HOLD with a clear description of what would resolve the uncertainty
    • source-index.md 378 B 
      # Source index

- **Source repository:** https://github.com/magnus919/hermes-profiles
- **Inspected commit:** `867a555`
- **Imported source directory:** `verification-methodology`
- **Porting boundary:** Retained portable methodology, templates, scripts, and references. Removed or generalized Hermes profile, task-orchestration, memory, and rigid response-handoff assumptions.
    • verdict-template.md 478 B 
      # Verdict Template

## Verdict

**PASS / HOLD / FAIL**

## Criteria Assessment

| Criterion | Result | Evidence | Notes |
|-----------|--------|----------|-------|
| Criterion 1 | PASS/FAIL/HOLD | What evidence supports this | Any caveats |

## Summary

- **Passed:** [count] criteria
- **Failed:** [count] criteria
- **Held:** [count] criteria (insufficient evidence)

## Next Steps

If HOLD: what would resolve the uncertainty.
If FAIL: what must change before re-submission.
  • README.md 2.1 KB 
    # Verification Methodology

Replace completion claims with a disciplined evidence trail that shows what was checked, what passed, and what remains uncertain.

## Why Install This Skill

Replace completion claims with a disciplined evidence trail that shows what was checked, what passed, and what remains uncertain. It preserves a practical method, local reference material, and reusable templates so an agent can do more than produce a generic answer.

Use it when the work needs a repeatable process and an inspectable result. It is portable across Agent Skills-compatible clients and does not require a profile system or a particular task orchestrator.

## What You Get

| Path | What it provides |
|---|---|
| `SKILL.md` | Trigger conditions, workflow, and guidance for loading deeper resources. |
| `references/` | Reference material: `criteria-assessment.md`, `evidence-standards.md`, `verdict-template.md` |
| `evals/evals.json` | Regression cases for direct-source selection, bundled executable discovery, failure reporting, and explicit fallback handling. |

## Quick Start

Define the criteria, collect evidence from the requested source before any substitute, and produce a verdict using `references/verdict-template.md`.

Install or expose this directory using your agent's standard Agent Skills loading mechanism, then ask for work that matches the triggers below.

## Triggers

- Verify work against explicit criteria using direct, source-faithful evidence, reproducible checks, and clear verdicts. Use before declaring an artifact, implementation, or claim complete.
- Requests involving the method, deliverables, or review process described in `SKILL.md`.
- Work where a reusable template or reference from this skill would reduce avoidable mistakes.

## Requirements

No runtime dependency.

## Source and maintenance

This skill was extracted from [`magnus919/hermes-profiles`](https://github.com/magnus919/hermes-profiles) at commit [`867a555`](https://github.com/magnus919/hermes-profiles/commit/867a555). The portable methodology was retained; Hermes-specific profile, orchestration, and memory assumptions were removed.
  • SKILL.md 8.3 KB 
    ---
name: verification-methodology
description: Verify work against explicit criteria using direct, source-faithful evidence, reproducible checks, and clear verdicts. Use before declaring an artifact, implementation, or claim complete; do not use for exploratory research without pass/fail criteria.
license: MIT
compatibility: No runtime dependency.
metadata:
  source_repo: https://github.com/magnus919/hermes-profiles
  source_commit: 867a555
---


# Verification Methodology

Pass/fail assessment against pre-defined criteria.

## The Verification Protocol

1. **Receive** — restate the artifact, claim, or implementation being verified and the decision it will support.
2. **Assess criteria** — convert requirements into observable pass/fail conditions; identify what would disprove each claim.
3. **Investigate** — collect direct, reproducible evidence from the source named by the request and record commands, source locations, or source URLs. Own this collection when the source is accessible: do not ask the user to relay evidence you can retrieve yourself. Ask only for access you genuinely lack.
4. **Decide** — mark each criterion passed, failed, blocked, or not applicable. Do not convert missing evidence into a pass.
   - When verification is split across lanes, no lane's PASS is the gate verdict. Wait for every required lane and reconcile conflicts before synthesis.
   - A later source-fidelity BLOCK supersedes an earlier prose, rendering, or checklist PASS for the same artifact hash. Immediately label earlier PASS packets as superseded so downstream workers cannot mistake them for the current verdict.
   - Pattern scans and internal consistency cannot substitute for checking factual transformations against the named primary source or source dossier.
5. **Report** — use the verdict template to distinguish verified facts, assumptions, and remaining work.

Stop when every criterion has direct evidence or an explicit blocked/not-applicable verdict. Escalate when the criterion is ambiguous, evidence conflicts, or the required access is unavailable.

## Source Fidelity

Treat the requested or configured local service as part of the verification criterion, not as an interchangeable topic label.

1. Load the matching skill and use its documented executable or service path before adjacent integrations, generic web search, or public project sources.
2. Use the path resolved by the skill itself. A missing global `PATH` entry does not prove that a bundled executable is unavailable.
3. If the direct source fails, report the attempted command or endpoint and its exact failure. Do not silently substitute evidence from another source.
4. Use a substitute only when the user requests broader context or explicitly accepts the fallback. Label substitute evidence as secondary and do not present it as the requested source's state.

Example: for “What’s new on Jellyfin?” in an environment with a configured Jellyfin skill and bundled CLI, query that server through the bundled CLI first. Home Assistant entities and public Jellyfin project activity answer different questions.

## When not to use

Do not use this skill for open-ended exploration that has no artifact, claim, decision, or observable completion criterion. Use a research or discovery skill first, then return here when there is something falsifiable to verify.

## Related Skills

- [playwright](../playwright/SKILL.md) — browser-based E2E checks as a verification instrument: authoring and running specs against the real UI boundary.
- [documents](../documents/SKILL.md) — producing verification reports or evidence packets as PDF/Word/Excel/PowerPoint deliverables.

## Reference Files

| Reference | When to load |
|-----------|-------------|
| `references/criteria-assessment.md` | You need to evaluate whether work meets completion criteria |
| `references/evidence-standards.md` | You need to judge whether evidence supports the claims made |
| `references/magnus919-refine-to-ship-gate.md` | You are running the Magnus919 Refine-to-Ship verifier gate — 12 criteria, editorial change verification, output structure |
| `references/verdict-template.md` | You need to produce a structured pass/fail/hold verdict |

## Magnus919 Refine-to-Ship Gate Criteria

12 criteria for the verifier profile. Each criterion maps to an observable, reproducible check.

### All 12 Criteria

| # | Criterion | How to Verify |
|---|-----------|---------------|
| 1 | **Dash scan** — zero em dash (U+2014), en dash (U+2013), horizontal bar (U+2015), or visible prose double-hyphen | `search_files` for `[\u2014\u2013\u2015]` and `\-\-`. Double-hyphens in YAML frontmatter delimiters are OK. |
| 2 | **Fact-check** — all methodology claims map to source; no fabricated numbers, chronology, or universal claims | Cross-reference article claims to source document sections. Search for `\d+%`, `percent`, `average of`, `illustrative`. Search for `research proves`, `studies demonstrate`. |
| 3 | **Voice-check** — Magnus fingerprint: conversational first-person, contractions, "But" pivots (not formal transitions), colons over semicolons, no consultant cadence | Search for `Furthermore`, `Moreover`, `Nevertheless`, `Consequently`, `Therefore`, `not only.*but also`, triplet parallelism. Count colons vs semicolons (should skew heavily toward colons). |
| 4 | **Oxford commas, spelling, grammar** — American English, Oxford commas in series, no spelling errors | Manual read of series. Check for consistent formatting. |
| 5 | **No formulaic AI closing** — zero "In conclusion", "To summarize", "In this article", generic motivational advice | `search_files` for `In conclusion`, `Ultimately,`, `To summarize`, `In this article`, `In this post`. |
| 6 | **Methodology-first** — personal frame ≤ ~10% of article; rest is methodology | Count paragraphs in frame vs body. |
| 7 | **Human stake integrated** — cognitive burden, expertise formation, transferred work, anti-surveillance, accountable authority | Verify dedicated section or dispersed coverage of all dimensions. |
| 8 | **Privacy/anonymization** — zero company identifiers, role titles, named people, source filename, proprietary domain examples | `search_files` for company name, product names, domain-specific terminology from source. |
| 9 | **Frontmatter** — title, slug, date, byline correct and value-identical to specification | `read_file` lines 1–11. |
| 10 | **Links resolve** — each distinct URL appears once at first meaningful mention; all return 200 | `curl -s -o /dev/null -w "%{http_code}"` each URL. Verify link text is at first meaningful mention. |
| 11 | **Hugo build + routes** — build exit 0; new route returns 200; old take-home-title route returns 404 | `hugo --quiet && echo EXIT:$?`. `curl` both routes. |
| 12 | **No duplicate source bundle** — single directory, single `index.md`; no stale `*take-home*` directories | `ls` the page bundle directory. `find` in content/posts for duplicate slug patterns. |

### Parent-Requested Editorial Changes

When the parent profile specifies editorial changes during gate recovery, verify each one is present before proceeding with the full criteria scan:

| Change Type | Verification Method |
|-------------|-------------------|
| Fabricated illustrative numbers removed | `search_files` for `\d+%`, `percent`, `PRs? per`, `average of` → zero hits |
| Tense correction | `search_files` for the exact parent-specified phrase |
| Closing replacement | `search_files` for the first and last sentence of the parent-specified closing |

### Verdict Rules

- **PASS:** All 12 criteria met. Produce 00-index.md, 01-summary/verdict.md, 02-analysis/per-criteria-results.md.
- **BLOCK:** Any criterion fails. Produce gap-details.md with specific fix instructions. See `verifier-gate-recovery` skill for remediation patterns.

### Output Structure

```
/private/tmp/verifier-gate/<slug>-refine/
  00-index.md              — verdict, links to artifacts
  01-summary/verdict.md    — per-criterion pass/fail table
  02-analysis/
    per-criteria-results.md  — detailed evidence per criterion
    gap-details.md           — only if BLOCK, with remediation instructions
```

## Portability

This skill is intentionally host-neutral. Use your agent's normal mechanisms to load the references, templates, and scripts listed here. Do not assume a particular profile system, task orchestrator, memory service, or response-handoff format.

Comments (0)

Sign in to join the conversation.

No comments yet.

Reviews (0)

No reviews yet.

Related

Claude Prompt
Tighten-my-prose editing prompt
Claude Skill
Adversarial code-review skill for Claude
Claude Workflow
Deep-research workflow (fan-out + verify)
Claude Skill
Conventional-commit writer for Claude