Tml hygiene

Skill CobraChickenAI/tml-capture-kit/skills/tml-hygiene

Capture architecture from conversation. One pipeline, seven agent skills, a plain-language map anyone can read.

Install

npx -y skills add CobraChickenAI/tml-capture-kit --skill tml-hygiene

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Quality-control pass for TML outputs. Reviews Map, Inferences, Changelog, So What, and optional Agent Brief for consistency, provenance completeness, ambiguity, and actionability before sharing. Part of the TML capture pipeline — runs last, after tml-so-what. Final gate before sharing with stakeholders.

The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

3.7 KB, 704 tokens by cl100k_base, as published. Nobody here has run it

TML Hygiene

Run this skill after a mapping session and before sharing outputs with stakeholders.

The goal is not to rewrite everything. The goal is to catch drift, contradictions, weak provenance, and vague recommendations while fixes are still cheap.

When To Use

After tml-capture completes
Before sending outputs to a manager, DPM, or external partner
During repo review, acceptance testing, and pilot rollouts

Inputs

Provide:

The Map
The Inferences
The Changelog
The So What
Optional Agent Brief

If any artifact is missing, flag it as a blocker.

Checks

1. Artifact Completeness

Confirm all required sections are present and non-empty.

Source registry exists (Source 1, Source 2, ...).
Map has: scope summary, what we do, who is involved, rules, connectors/bindings, provenance section.
Inferences has: stated vs inferred breakdown and confidence.
Changelog has: source references and status per major element.
So What has: visibility gaps, friction points, coherence risks, quick wins, AI leverage points, single recommended starting point.

2. Cross-Artifact Consistency

Find contradictions between documents.

Names match across artifacts for scope/domains/capabilities.
Policies in map match policy references in inferences/changelog.
So What findings point to elements that actually exist in the map.
Agent Brief permissions match map policies and archetypes.

3. Provenance Quality

Check whether claims trace to a source.

Every artifact references source ids from the registry.
Every major map element has a source path in changelog.
Inference claims distinguish stated vs inferred.
Low-confidence items are explicit and reviewable.
No unsupported “fact” appears without a source or inference label.

4. Actionability

Reject generic recommendations.

Quick wins include a concrete action, estimated effort, and expected impact.
AI leverage points define operating archetype, policies, read/write boundaries, and explicit cannot-do constraints.
Recommended starting point is singular, specific, and feasible this week.

5. Language Hygiene

Ensure outputs are clear for non-technical stakeholders.

Plain language, short sentences, minimal jargon.
No vague filler (“optimize,” “enhance,” “leverage synergies”) without specifics.
Uncertainty is explicit, not hidden.

Output

Return one hygiene report:

# Hygiene Report: [Scope Name]

## Status
- PASS | PASS WITH FIXES | BLOCKED

## Findings
1. [Severity: High/Med/Low] [Issue]
   - Where: [artifact + section]
   - Why it matters: [impact]
   - Fix: [specific edit]

## Missing Inputs
- [any missing artifacts]

## Fast Fixes (Do Now)
1. [highest impact quick fix]
2. [next fix]
3. [next fix]

## Ready To Share
- Yes/No
- If no, what remains:
  - [list]

Rules

Be direct and concrete.
Do not invent missing facts; request correction.
Prioritize trust issues first (provenance, contradictions, permissions).
Keep final recommendations implementation-ready.

v1.0 — quality gate for release readiness and stakeholder confidence.