wassname/ml-debug
1
0
Fork 0
mirror of https://github.com/wassname/ml-debug.git synced 2026-09-09 11:27:03 +08:00
Code Issues Packages Projects Releases Wiki Activity

skill: bug-assumption gets its own trigger, the all-clear sentence; guide cap 400 -> 180 lines

Browse Source 
A sonnet subagent read the skill, summarised the ritual list, ran none of them,
and closed with 'everything checks out'. Phase triggers do not fire; a sentence
the agent watches itself write might.

Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
This commit is contained in:
wassname
2026-08-19 20:44:49 +08:00
co-authored by Claudypoo
parent d37b9c88a6
commit 4f422a8016
1 changed files with 16 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files
SKILL.md
+16 -3
View File
@@ -21,6 +21,18 @@ english. Use these words for every credence, from the
|---|---|---|---|---|---|---|
| 86-99% | 71-85% | 56-70% | 46-55% | 31-45% | 16-30% | 1-15% |
## 0. You have a bug. The trigger is the all-clear
This one has no phase, so watch your own draft instead. The moment you are about to write
"everything checks out", "nothing looks unresolved", "no open questions", "looks good", or "the
metrics look fine", stop and pay ritual 6 first: three diagnoses with credences, five ways this is
invalid, and random raw samples.
An all-clear is a claim, and it is the claim least likely to have been checked. A machine learning
system has many adaptive parts, so a broken one is often hidden by the others compensating, and the
output still looks reasonable. Raise the bar at which you say "I think this is correct". Most
research results are false, and an exciting result is more likely false than a boring one.
## 1. Frame the project, once, before the first real run
Write this into the project's `AGENTS.md` and keep it there. Without it, a null result later gets
@@ -38,10 +50,11 @@ EVIDENCE B, quantitative: beats a named baseline on a named metric, not chance.
ELSE: if A and B disagree, or either fails, say now what that means about the method.
```
## 2. Keep a training guide, under 400 lines, living
## 2. Keep a training guide, two pages, living
`TRAINING_GUIDE.md` in the project. The worker maintains it. It holds the current model of the
system, not a run history, so it stays small by replacing claims that went stale.
`TRAINING_GUIDE.md` in the project, under 180 lines. The worker maintains it. It holds the current
model of the system, not a run history, so it stays small by replacing claims that went stale. Past
180 lines you are appending history, and nobody will read it, which defeats the file.
Frontmatter carries only what a program can check: