mcpbeat

Deliverable Selfloop

willoscar/deliverable-selfloop

Score an existing reader-facing deliverable and emit a deterministic PASS/FAIL scorecard plus bounded repair plan; use after synthesis, not to rewrite content.

4k tokens
context cost
the whole folder, loaded on every use
4
files
ships runnable scripts
0
copies elsewhere
how many repositories repackaged it
496
stars on the repo
on the repository, not the skill itself

Install

one command, takes just this skill from the repository 
npx skills add https://github.com/WILLOSCAR/research-units-pipeline-skills --skill deliverable-selfloop

What comes with it

12 492 bytes besides the instruction
references/overview.md
references/quality_checklist.md
scripts/run.py

The instruction itself

7 sections, as written by the author

Deliverable Self-Loop

Runs the final quality gate for a reader-facing deliverable and always writes a PASS/FAIL report.

Inputs

Primary input depends on the active pipeline contract:

  • brief deliverable
  • paper-review deliverable
  • evidence-review deliverable
  • idea memo bundle
  • tutorial deliverable

Output

  • output/DELIVERABLE_SELFLOOP_TODO.md
  • for research-brief: output/BRIEF_SCORECARD.md and output/BRIEF_SCORECARD.json
  • for paper-review: output/REVIEW_SCORECARD.md and output/REVIEW_SCORECARD.json
  • for idea-brainstorm: output/IDEA_SCORECARD.md and output/IDEA_SCORECARD.json
  • for evidence-review: output/EVIDENCE_SCORECARD.md and output/EVIDENCE_SCORECARD.json

Dispatch rule

The gate should dispatch by pipeline contract first:

  • quality_contract.deliverable_kind

Only fall back to legacy profile-name checks when contract metadata is missing.

Script boundary

scripts/run.py should:

  • detect the active deliverable contract
  • run the matching evaluator
  • always write a report

It should not mutate the deliverable itself.

Acceptance

  • report exists
  • report contains - Status: PASS or - Status: FAIL
  • PASS only when the active deliverable satisfies its minimum section / artifact contract
  • research-brief additionally requires valid core-set pointers and the configured score threshold
  • paper-review additionally requires the configured score threshold and all critical rubric dimensions
  • idea-brainstorm additionally requires traceable anchors, actionable lead directions, and the configured score threshold
  • evidence-review additionally requires clause-linked screening, complete extraction rows, synthesis pointers, and the configured score threshold

Non-goals

  • rewriting the deliverable
  • choosing upstream fixes beyond pointing to the missing contract items

How to use it

Copy the folder

Take willoscar/deliverable-selfloop from the repository into ~/.claude/skills for personal use, or into .claude/skills inside a project.

Check the name does not clash

The agent identifies a skill by the name field in its header. Two skills with the same name cannot sit side by side — one of them will be ignored.