agentleFS
link7373link7373/agentic-bi-team/.claude/skills/make-deliverable/SKILL.md

make-deliverable

Skill4 starsChanged 6 days agoMITView on GitHub

Produce a stakeholder-ready deliverable from completed analysis — PowerPoint/Google Slides deck, Word/Google Doc report, or Excel/Google Sheets workbook — with pyramid structure, takeaway titles, and every figure traceable to source. Use for any output going to humans beyond the team.

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

2 rules. About 0 are common, 1 is distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Google Workspace native: if access is configured ({{GOOGLE_WORKSPACE_ACCESS or "not configured"}}), create directly.

    line 20
  2. Distinctive:

    Governance gate: Run the output scan on the folder, including the rendered Office files:

    line 28

≈ “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

Make Deliverable

Owner: insights-communicator. Args name the source analysis and format, e.g. /make-deliverable Q2 churn analysis as exec deck.

Procedure

  1. Identify source and audience. Source: a FINDINGS.md (or scorecard/model card) — if the analysis doesn't exist yet, run /analyze first; this skill packages, it doesn't analyse. Audience: from knowledge/stakeholders.md — their altitude, preferences, and format default ({{DELIVERABLE_FORMATS}}).
  2. Outline the storyline (pyramid principle): governing answer → 2–4 supporting points → evidence. Every section title is a full-sentence takeaway. For board/external/high-stakes audiences, show the user the outline before full production.
  3. Produce the artifact into deliverables/YYYY-MM-DD-<slug>/:
    • PowerPoint: python-pptx. Slide 1 = executive summary that stands alone. One idea per slide; takeaway-sentence titles; charts via matplotlib in the standard palette (standards/reporting-standards.md), embedded as images.
    • Word/Docs: python-docx. Executive summary ≤1 page → findings → recommendation → methodology appendix.
    • Excel/Sheets: openpyxl. Sheet 1 = formatted Summary; data sheets with frozen headers and proper number formats; final Notes sheet (sources, refresh date, definitions).
    • Google Workspace native: if access is configured ({{GOOGLE_WORKSPACE_ACCESS or "not configured"}}), create directly; else deliver the Office-format file ready to upload.
    • Markdown/PDF: for internal audiences; convert when asked. Install needed Python packages if absent.
  4. Integrity pass. Create SOURCES.md mapping every key figure in the deliverable to its query/analysis file. No number without a source. Carry the source analysis's provenance footer into the deliverable (a footnote on the summary page, or the Notes sheet): a reader of the deck should be able to see the source tier and the data-through date without opening the repo. Caveats included proportionately — footnote-level unless a caveat changes the conclusion (then it's a headline).
  5. Quality pass. Honest axes (bars from zero), consistent rounding and precision, comparisons anchored ("vs Q1", "vs target"), brand/format rules applied ({{BRAND_GUIDELINES or "clean defaults"}}), the first page passes the stand-alone test.
  6. Governance gate. Run the output scan on the folder, including the rendered Office files:

    python scripts/scan_outputs.py deliverables/YYYY-MM-DD-<slug>

    Fix every finding in the source and regenerate — an email in an appendix table, a breakdown with a group under the minimum size (suppress it, and its complement if it can be recovered by subtraction). A value that is legitimately fine goes on the allowlist in knowledge/governance-policy.md, with a reason. Check the audience against each metric's classification in the catalog: a confidential metric goes only to its named audience. A clean scan does not find a name in free text — read the artifact too.

  7. Review. Anything for leadership, a board, or anyone outside the company goes to analysis-reviewer before delivery, with the source analysis. It checks the deliverable against the analysis, not just the analysis: a number that was right in FINDINGS.md and rounded, relabelled, or stripped of its caveat on a slide is the most common defect at this stage.
  8. Deliver. Send the file(s) to the user. Distribution beyond the user (email, shared drives, external parties) is outward-facing — confirm first unless pre-authorised in CLAUDE.md §8.

Created by Colin Beck — https://www.linkedin.com/in/beckcolin/

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.