agentleFS
mblaubergmblauberg/provenant/skills/deliver/SKILL.md

deliver

Skill1 starsChanged 5 days agoMITView on GitHub

Use to verify and accept approved non-software outcomes (research, analysis, documents, agent work) and to write the outcome report every skill uses. Not for software changes (implement), unsettled scope (scope) or promotion (release).

25 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use prune, create-issue and 8 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • matthewpSkill

    prune

    Find and remove unused CSS in a Tideway project — classes defined in stylesheets but never referenced in markup. Use whenever the user wants to prune dead CSS, find unused styles or classes, clean up or slim a stylesheet, find orphaned BEM classes, or "tree-shake" hand-written CSS.

    Also in 3 projects that use this

  • sahinSkill

    create-issue

    Also in 3 projects that use this

  • BlockSkill

    plan

    Draft a TDD, PRD, or lightweight plan in the conversation without committing it.

    Also in 5 projects that use this

  • SentrySkill

    spec

    Produce a verified spec — problem, desired outcome, and acceptance criteria — before building. Use when given a GitHub/Linear issue link, when starting from a fuzzy in-conversation idea, or asked to "figure out what to do" / "scope this" / "what should I build" — whenever the *what* isn't yet pinned down. Grills the ambiguities, checks the premise against the code, and produces acceptance criteria you sign off on — then hands to design-first.

    Also in 4 projects that use this

  • AWSSkill

    ship

    Finalize dev work on an issue branch — update docs, push, and open a PR.

    Also in 5 projects that use this

  • AWSSkill

    wait-what

    Stop. That last message did not land: re-pitch it.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

4 rules. 0 are common, 2 are distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 25 edited versions we read keep these.

  • 2

    Distinctive

    This author's own choices. Worth a look.

  • 2

    Very specific

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

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

    Before starting an issue, show tracker lint advice and proceed even if checks are unavailable.

    line 21
  2. Very specific:

    Agent work products: check the output against the criteria directly.

    line 33
  3. Distinctive:

    Follow-ups, each as a tracker issue or a named owner.

    line 47
  4. Distinctive:

    Never fabricate approval from passing checks.

    line 49

“Common” is measured: kept by at least 6 in 10 of the 25 edited versions of this file we could read. “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

Deliver

Non-software work has no CI to say it is done. This skill supplies the evidence standard and the report that lets the owner accept or reject the result without redoing it. Software goes through implement, which ends with the same report.

Entry

Require approved intent, acceptance criteria with how each is verified, a risk tier and bounded authority. Unsettled intent goes to scope. A run that needs a delivery record saves one with provenant workflow receipt save ID --expected-revision N --request FILE and reads it with provenant workflow receipt show ID.

Before starting an issue, show tracker lint advice and proceed even if checks are unavailable. provenant run --use deliver shows it on entry; otherwise run provenant tracker lint N (or --file PATH for a local spec).

How it runs

  • Research or analysis: provenant run --use research --file Q.md --explain to gather sourced claims, then synthesise to reconcile them. Every claim carries a source, a date and whether it was observed or inferred.
  • Documents: a read-only fidelity check through run --use verify, preferably on a different model family, that compares the document with its sources and criteria and lists unsupported claims, not stylistic edits.
  • Agent work products: check the output against the criteria directly; produce deterministic evidence (a diff, a recomputation, a test) before judgement evidence. For stochastic behaviour use evaluate.

Outcome report

The owner can reject it, so write it for a reader who will not open the artefact. Lead with the decision needed, then:

  1. What changed or was produced, and where it lives.
  2. Each acceptance criterion with its evidence, or unverified and why.
  3. Review coverage: who or what reviewed, from which model family, and what was not covered.
  4. What was not verified, and the assumptions made on the owner's behalf.
  5. Follow-ups, each as a tracker issue or a named owner.

Never fabricate approval from passing checks. A check that did not run is reported as not run. Evidence outranks confidence.

Acceptance

Objective checks plus the review the tier requires meet acceptance; the chair accepts and reports, and the owner may reject. Explicit owner acceptance is needed only for terminal work or when the approved scope asks for it. Delegates may only narrow authority. Sending, publishing and deploying are release. Keep the criteria, evidence and outcome in the existing issue or PR, and move settled rules to their durable home (engineering-docs).

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.