verdict
Skill1 starsChanged 4 months agoMITView on GitHub
Append-only judgment ledger. Every DOOR / GATE / PREMORTEM / REFUTE / DRIFT verdict line gets logged to one grep-able file so "show me every gate we overrode" is a query, not archaeology. Use whenever a judgment skill produces a verdict, and to query past verdicts.
10 other projects use a version of this.
Adopted by
Most recently taken up by
- rhasselle-oddball/claude-skillstheir own version, first seen Nov 1, 2026
- duthaho/skillhubtheir own version, first seen Oct 31, 2026
- udecode/platetheir own version, first seen Oct 17, 2026
- alexjmiller5/agent-config-publictheir own version, first seen Oct 15, 2026
- jacobgreenwald21/claude-skillstheir own version, first seen Oct 15, 2026
- arbelh/perduretheir own version, first seen Oct 11, 2026
Other versions10 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 10
- udecode/plate★ 17ksame name
- SanketSapkal/founder-skills★ 10same name
- damionrashford/Adversaria★ 9same name
- duthaho/skillhub★ 9same name
- alexjmiller5/agent-config-public★ 0same name
- andrewcovenant/skills★ 0same name
- arbelh/perdure★ 0same name
- jacobgreenwald21/claude-skills★ 0same name
- OpentypeAI/opentype-claude-plugin★ 0same name
- rhasselle-oddball/claude-skills★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
The file
We split this skill into the rules it gives an agent
3 rules. 0 are common, 1 is distinctive, and 2 are very specific to this project.
0
Common
Most of the 9 edited versions we read keep these.
1
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
- Very specific: line 38
MUST name a human in by: — a model cannot accept risk on anyone's behalf.
- Distinctive: line 55
If a project keeps decisions elsewhere (e.g.
- Very specific: line 60
Every judgment skill (/door /gate /premortem /refute /drift /altitude /think /sweep) — they emit, this records.
“Common” is measured: kept by at least 6 in 10 of the 9 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.
Files
/verdict — The Judgment Ledger
Decisions evaporate; ledgers compound. Every verdict the judgment skills produce is one line here. Six months later the ledger answers: what did we lock in, what did we override, what did we accept, and who signed it.
Invocation
/verdict log DOOR: one-way | locked: uuid pks | escape: full remap | gate: join p95
/verdict show 20 # last 20 verdicts
/verdict grep GATE # all gate verdicts (or any pattern)
/verdict stats # counts by type, overrides, accepted risksLedger location & format
File: .claude/verdicts.log in the project (create on first log; add to the
repo, NOT gitignore — the ledger is the point). One line per verdict:
2026-07-09 | DOOR | one-way | locked: uuid pks | escape: full remap | by: operator
2026-07-09 | GATE | TRIPPED p95=340ms>200ms | on-fail: STOP executed | by: claude
2026-07-10 | GATE | OVERRIDE 200→400ms | reason: cold cache accepted | by: operator
2026-07-11 | REFUTE | REFUTED "search works" | broke on: unicode names | by: claudeFields: date | TYPE | verdict | detail… | by: <who>. TYPE is one of
DOOR, GATE, PREMORTEM, REFUTE, DRIFT, ALTITUDE. Overrides and risk-acceptances
MUST name a human in by: — a model cannot accept risk on anyone's behalf.
Procedure
- log — append the line verbatim (plus date + by). Never edit or delete prior lines; corrections are new lines referencing the old.
- show / grep — read and filter; render as-is.
- stats — counts per TYPE, number of GATE OVERRIDE lines, number of PREMORTEM accepted-risks, oldest unresolved one-way door. Flag anything suspicious plainly (e.g. "4 gate overrides in 2 weeks, all same gate — the threshold is wrong or the gate is being gamed").
Rules
- A verdict not logged is a verdict that didn't happen. The judgment skills end by calling this.
- The ledger is append-only. History that can be rewritten is not history.
- If a project keeps decisions elsewhere (e.g. a decisions DB), log there TOO, but the flat file stays — it survives every infra migration.
Composes with
- Every judgment skill (
/door /gate /premortem /refute /drift /altitude /think /sweep) — they emit, this records. /calibrate— reads this ledger to score how past confidence aged./precedent— greps this ledger as its primary store of prior rulings./escalate— every queue adjudication lands here; the ledger feeds future precedent.
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.