agentleFS
iwesysiwesys/IWE/.claude/skills/vdv/SKILL.md

vdv

Skill0 starsChanged 25 days agoMITView on GitHub

The VDV Skill is a generator and auditor for staged process descriptions

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 5 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read the activity description from $ARGUMENTS.

    line 72
  2. Distinctive:

    After the VDV table, output a compact block:

    line 100
  3. Distinctive:

    After user edits — run a full audit (steps 1–5 of audit mode) including principles 5–6.

    line 115
  4. Distinctive:

    If the user disagrees — clarify.

    line 133
  5. Distinctive:

    If there are ❌ violations — propose a corrected stage table with fixes applied.

    line 170

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

VDV Skill (Input·Action·Output)

Service Clause: DP.SC.052
Source of principles: PD.METHOD.008 §174-230 (strategy cascade v9 as reference test case)

Execute: $ARGUMENTS


When to use

The VDV Skill is a generator and auditor for staged process descriptions based on the 6 Input·Action·Output principles. Use it to build a new process description (/vdv build) or to audit a ready-made description (/vdv audit).

VDV Principles (audit reference)

Six invariants used in both modes:

#PrincipleStatementViolation test
1Triad at every stageEach stage has an Input, an Action, and an Output. Missing one means incomplete.One of the three fields is absent or empty
2Output→Input chainingThe output of a stage must become the input of at least one subsequent stage.The output of stage N does not appear in the Input of any later stage (and is not marked (external) / (perimeter promise))
3Input invariantInput = outputs of preceding stages + stable documents + "plan for comparison". An artifact with no producer and no (external) label signals a missing stage.The Input of stage N contains an artifact that is not in the Output of any preceding stage and carries no (external) label
4Bidirectional traceabilityEvery artifact has a producer and a consumer. No consumer — redundant. No producer and no (external) — missing stage.An artifact in an Output has no consumer in the Inputs of subsequent stages (and is not a perimeter promise)
5Carry-overIncomplete work carries over into the Input of the next iteration of the same perimeter.Repeating process (rhythm > 1 iteration): the input of the first stage does not include carry-over from previous iteration
6Anti-dummyA stage has direct or indirect traceability to the perimeter promise. Remove the stage — does the promise break? If not — dummy candidate.Check: remove the stage — does the perimeter promise break? No → ❌ dummy candidate. The stage traces neither directly (output = promise) nor indirectly (output chain leads to promise)

Special cases for principle 5: single-iteration process (one-shot, no cycle) — principle 5 does not apply; mark ⏸ with explanation.
Special cases for principle 6: perimeter promise is unknown — ⚠️ + request clarification.


Algorithm

Mode /vdv build — Generation

Algorithm (approach C)

Step 1. Understand the activity

Read the activity description from $ARGUMENTS. If the description is too brief (fewer than 1 sentence, or no result/output is mentioned) — request clarification:

"Please describe the activity in more detail: what happens, what is the main result, is there a repeating rhythm?"

Step 2. Extract stages (quick draft)

Derive a tentative set of stages from the description. Rules:

  • Each stage = one semantic action with a verifiable artifact as output
  • First stage: mark inputs as (external) if they are not produced inside the process
  • Specify rhythm if known from context, otherwise —
  • Format: compact markdown table, columns strictly in order: # | Stage | Rhythm | Input | Action | Output

Step 3. Self-audit principles 1–4

Immediately after the draft, run principles 1–4 against the built table. Show:

Preliminary audit (principles 1–4):
P1 (triad): ✅ / ⚠️ / ❌ — <findings>
P2 (chaining): ✅ / ⚠️ / ❌ — <findings>
P3 (input invariant): ✅ / ⚠️ / ❌ — <findings>
P4 (traceability): ✅ / ⚠️ / ❌ — <findings>
P5 (carry-over): ⏸ will be checked after structure is confirmed
P6 (anti-dummy): ⏸ will be checked after perimeter promise is confirmed

Step 4. Inline traceability

After the VDV table, output a compact block:

| Artifact | Produced at stage | Consumed at stages |
|----------|-------------------|--------------------|
| Name     | N. Stage          | M. Stage / → perimeter promise / → carry-over (next iteration) |

Dangling artifacts (no consumer inside the perimeter) are labeled:

  • → perimeter promise — terminal output, serves as the promise
  • → carry-over (next iteration) — flows into the next process cycle (P5)
  • → ❌ consumer not found — violation of P4

Step 5. Clarification and final audit

After user edits — run a full audit (steps 1–5 of audit mode) including principles 5–6.


Mode /vdv audit — Audit

Algorithm

Step 1. Receive the description

Input text from $ARGUMENTS — a stage table or a text description. If it does not contain an explicit stage structure — ask: "Please identify the stages or provide a table with Input/Action/Output columns."

Step 2. Derive the perimeter promise

Find all outputs that are not the input of any subsequent stage = candidates for the perimeter promise. Show:

"I see the perimeter promise as: <list of terminal outputs>. Is that correct?"

If the user disagrees — clarify. If there are multiple terminal outputs — ask the user to select or confirm all of them.

Step 3. Run the 6 principles

For each stage and each artifact, check all 6 principles. Verdict format:

## Process audit: <name>

### Principle 1: Triad at every stage
✅ All stages have a triad / ❌ Violations:
- Stage N "<name>": missing <Input | Action | Output>
  Fix: <specific suggestion>

### Principle 2: Output→Input chaining
...

### Principle 3: Input invariant
...

### Principle 4: Bidirectional traceability
...

### Principle 5: Carry-over
✅ / ⚠️ / ❌ / ⏸ <explanation>

### Principle 6: Anti-dummy
Perimeter promise: <what was accepted>
✅ / ⚠️ / ❌ — <for each ❌ stage: traceability to promise not found>
  Fix: <remove stage or explicitly link to promise>

---
Summary: ✅ <N principles OK> | ⚠️ <M warnings> | ❌ <K violations>

Step 4. Optional: propose a fix

If there are ❌ violations — propose a corrected stage table with fixes applied.


Reference /vdv

VDV Skill (Input·Action·Output) — WP-413, DP.SC.052

Modes:
  /vdv build <activity description>  — build a process description
  /vdv audit <stage description>      — audit against 6 VDV principles

6 principles: triad · chaining · input invariant · traceability · carry-over · anti-dummy

Reference test case: strategy cascade v9 (PD.METHOD.008 §185-189)
Development test cases: test_cases.md
<!-- USER-SPACE --><!-- /USER-SPACE -->

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.