verify-the-effect
Skill27 starsChanged 18 days agoApache-2.0View on GitHub
Playbook for the always-on norm "never report success from a proxy signal — assert the actual effect or artifact". Load when a status signal (green CI step, HTTP 200, "accepted", exit 0, progress bar) is about to be read as success, when a step's entire output is a side effect, or when routed here from the one-real-ride done gate. Does not fire on test-suite results (prove-it-can-fail) or on declaring a whole feature done (one-real-ride owns that moment).
In this repository56 other files it gives its agents.
- ai-surface-disciplineSkill
- blast-radius-grepSkill
- checkpoint-logSkill
- commission-the-rosterSkill
- committee-reviewSkill
- confirm-before-pushSkill
- constant-time-secret-compareSkill
- context-economySkill
- contract-reviewSkill
- cross-agent-reviewSkill
- date-parse-utc-safeSkill
- db-migration-verificationSkill
- deliverable-integritySkill
- dependency-upgrade-verificationSkill
- earn-every-lineSkill
- email-delivery-verificationSkill
- enforce-invariants-in-buildSkill
- env-change-verificationSkill
- findings-are-evidenceSkill
- fixtures-from-realitySkill
- flag-deferred-itemsSkill
- guard-the-spend-pathsSkill
- handoverSkill
- honest-failure-surfacingSkill
- legal-citation-integritySkill
- legal-notice-handlingSkill
- live-data-surgerySkill
- live-state-firstSkill
- lock-at-the-chokepointSkill
- mass-red-triageSkill
- measure-per-routeSkill
- no-silent-data-dropSkill
- one-real-rideSkill
- outbound-side-effect-idempotencySkill
- parallel-work-reconSkill
- pin-to-what-was-seenSkill
- plan-firstSkill
- price-the-spendSkill
- prove-it-can-failSkill
- reproduce-the-real-buildSkill
- rerun-before-verdictSkill
- safe-smokesSkill
- secrets-in-outputSkill
- server-side-authoritySkill
- skill-library-builderSkill
- spawn-from-declared-stateSkill
- statute-to-obligations-registerSkill
- substantiate-outward-claimsSkill
- team-loopSkill
- tool-layer-bytesSkill
- trace-one-recordSkill
- unordered-eventsSkill
- unslop-codeSkill
- unslop-textSkill
- unslop-uiSkill
- verified-citationsSkill
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
- Distinctive: line 48
Every proxy signal exists because it is cheaper than the real assertion, and each incident in the corpus was invisible precisely as long as …
≈ “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.
Files
Verify the effect
The most recurrent failure in the lessons corpus (~nine ASH incidents plus the ICC email family): a success indicator that was true while the effect was absent. A green CI signing step that signed nothing. A report that "generated and emailed" while silently missing five sections. An HTTP 200 password change that never changed the password. "Accepted by Resend" masking a bounce. A served Content-Type differing from the uploaded one. Layer: norm + playbook — the one-line norm in the global CLAUDE.md is the everyday trigger; this file is the depth behind it.
The rule
A success indicator that can be true while the effect is absent proves nothing. Identify the real intended effect of the operation — the file signed, the row written, the email delivered, the section rendered, the password actually changed — and assert that.
How to apply
- Side-effect steps assert the side effect. Any step whose entire job is a side effect (signing, uploading, sending, publishing) gets an explicit post-step assertion: list the signature, HEAD the URL, read the delivery status. "The step was green" is not evidence — a Codemagic signing step once succeeded having signed nothing.
- Verify the artifact the user receives, not the pipeline's account of it: count its sections, count its photos, open the download link, check the served header rather than the uploaded one. A report can generate, email, and be missing half its content.
- Sensitive mutations are verified by using the result — sign in with the new password, read back the changed row — never by the response code. A WebView once returned 200 for a password change that silently no-opped.
- Diagnose from ground truth before theorising: server and auth logs, breadcrumbs, CI step durations (a two-second step that must call a remote API didn't call it), the running database, live object definitions. A working login proves only that one backend is reachable.
- Acceptance is not delivery. A provider 2xx means queued. For transactional email, load email-delivery-verification — the provider-specific child of this rule.
What this skill does not do
It does not gate feature completion — one-real-ride owns the "done/shipped" moment and routes here. It does not judge test results — prove-it-can-fail owns check integrity.
Why
Every proxy signal exists because it is cheaper than the real assertion, and each incident in the corpus was invisible precisely as long as the proxy was trusted. Evidence: ASH LESSONS_LEARNED 2026-05-21 (three incidents), 2026-06-04, 2026-06-07, 2026-06-09, 2026-07-14; ICC L-004/L-015 (shared evidence with email-delivery-verification).
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
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.

