agentleFS
randommoniclerandommonicle/claude-skills/no-silent-data-drop/SKILL.md

no-silent-data-drop

Skill27 starsChanged 18 days agoApache-2.0View on GitHub

Playbook for the always-on norm "filters and gates hide empty things; they never drop content". Load when adding any filter, gate, continue, catch, or fallback to a rendering or export path, when parsing model output that carries content, or when two renderers must produce the same content. Does not fire on error-message wording or telemetry levels (honest-failure-surfacing) or on filters over data the user never receives.

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 1 is distinctive, and 0 are 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.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    An ignore rule is a filter too: Before the first commit of new files, compare git status with the files written and run git check-ignore -v …

    line 52

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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

No silent data drop

The ASH 2026-06-04 cluster: user and model content silently vanishing from delivered reports through render gates, model-assigned routing keys, and JSON parse failures — discovered by the recipient, never by the pipeline. Layer: norm + playbook — the one-line norm in the global CLAUDE.md is the everyday trigger; this file is the depth.

The rule

Wherever code can discard data — a filter, a gate, a continue, a catch, a ?? fallback — ask: what happens to the non-matching data? A gate may hide an empty section; it must never drop content. Anything holding content still renders somewhere, and a safe default bucket always beats a drop.

How to apply

  • Gates hide empty things only. A property-flag render gate once dropped mis-classified photos from delivered reports for weeks; the fix routes non-matching content to a visible default section instead.
  • A parse failure never discards content. A truncated model JSON threw in JSON.parse and five report sections vanished silently. Salvage what can be salvaged, log the stop reason, surface the failure.
  • Model-emitted enums, keys, and ids used for routing are untrusted input. Clamp to the canonical vocabulary where the value enters AND again at render (stored rows may hold unclamped values), defaulting unknowns to a visible bucket.
  • Two renderers share the grouping function. Hand-synced twins drift, and the divergence masks the loss — one representation kept all photos while the other dropped them, so the audit trail looked complete.
  • Verify the delivered artifact (count sections, count photos), and be suspicious of any test that asserts content IS dropped — a passing test once encoded the bug.

What this skill does not do

It does not govern how failures are messaged to the user — that is honest-failure-surfacing. It does not apply to filters over data the user never receives (internal telemetry, debug output).

Why

Dropped content is the worst failure class in a document pipeline because the artifact still ships: everything looks successful, and only the recipient can notice the hole. Evidence: ASH LESSONS_LEARNED 2026-06-04 (all entries) and the CLAUDE.md max_tokens truncation lesson; ICC L-008.

Additions from the misses-log sweep (2026-10-22)

  • An ignore rule is a filter too. Before the first commit of new files, compare git status with the files written and run git check-ignore -v on any that is missing; a broad ignore pattern can hide a file meant to ship. (ICC L-101.)

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.