skills/boshu2__agentops
1
0
Fork 0
mirror of https://github.com/boshu2/agentops.git synced 2026-09-14 15:08:13 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix(learn): cite the living quarantine rule instead of paraphrasing a pruned incident

Browse Source 
Cross-family W5 review: pruning the dead verdict digests while keeping
the incident as an uncited paraphrase contradicted learn's own
pruned-not-paraphrased decay rule. The harvest example now cites the
mutating-check quarantine in skills/validate/SKILL.md — resolvable
lineage for the same incident.
This commit is contained in:
Bo
2026-07-28 10:01:35 -04:00
parent 529b48b4e3
commit 40339a81d9
5 changed files with 13 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files
images/gemini/skills/learn/SKILL.md
+3 -2
View File
@@ -41,8 +41,9 @@ size, and stop at advisory evidence.
Overweight failures: a `NOT_PROVEN` or `FAIL` verdict carries more teaching
value than a PASS, because it names a rule the loop lacked. Harvest kernels
from failed lanes first — a subject destroyed by a mutating check
mid-validation teaches a more durable rule than the PASS that follows it.
from failed lanes first — the canonical example is the mutating-check
quarantine in `skills/validate/SKILL.md`, a durable rule minted from a
`NOT_PROVEN`-then-`PASS` verdict pair.
Prune for provenance decay: every cited artifact must still resolve — the
file exists or the verdict digest is present under `.agents/ao/verdicts/`. A
skills-codex/.agentops-manifest.json
+2 -2
View File
@@ -453,8 +453,8 @@
{
"name": "learn",
"source_skill": "skills/learn",
"source_hash": "0a6d4e2c79acf1a7de1fd99def116ce7409ad45510cacc2bee5948f2223e6368",
"generated_hash": "bc43e0a2cfc0f22838d594eed769e24e0f582b31eb8726d07aa4fed6cfb9b457"
"source_hash": "53067b9717740db5e7b7c4f116e5c728f48893556789020c602a02a53723683c",
"generated_hash": "2cdb8ec214c209920dd724a2efd9dc5b2889835beb90e6490f315a69a5a22813"
},
{
"name": "ms",
skills-codex/learn/.agentops-generated.json
+2 -2
View File
@@ -2,6 +2,6 @@
"generator": "codex-sync",
"source_skill": "skills/learn",
"layout": "modular",
"source_hash": "0a6d4e2c79acf1a7de1fd99def116ce7409ad45510cacc2bee5948f2223e6368",
"generated_hash": "bc43e0a2cfc0f22838d594eed769e24e0f582b31eb8726d07aa4fed6cfb9b457"
"source_hash": "53067b9717740db5e7b7c4f116e5c728f48893556789020c602a02a53723683c",
"generated_hash": "2cdb8ec214c209920dd724a2efd9dc5b2889835beb90e6490f315a69a5a22813"
}
skills-codex/learn/SKILL.md
+3 -2
View File
@@ -18,8 +18,9 @@ size, and stop at advisory evidence.
Overweight failures: a `NOT_PROVEN` or `FAIL` verdict carries more teaching
value than a PASS, because it names a rule the loop lacked. Harvest kernels
from failed lanes first — a subject destroyed by a mutating check
mid-validation teaches a more durable rule than the PASS that follows it.
from failed lanes first — the canonical example is the mutating-check
quarantine in `skills/validate/SKILL.md`, a durable rule minted from a
`NOT_PROVEN`-then-`PASS` verdict pair.
Prune for provenance decay: every cited artifact must still resolve — the
file exists or the verdict digest is present under `.agents/ao/verdicts/`. A
skills/learn/SKILL.md
+3 -2
View File
@@ -41,8 +41,9 @@ size, and stop at advisory evidence.
Overweight failures: a `NOT_PROVEN` or `FAIL` verdict carries more teaching
value than a PASS, because it names a rule the loop lacked. Harvest kernels
from failed lanes first — a subject destroyed by a mutating check
mid-validation teaches a more durable rule than the PASS that follows it.
from failed lanes first — the canonical example is the mutating-check
quarantine in `skills/validate/SKILL.md`, a durable rule minted from a
`NOT_PROVEN`-then-`PASS` verdict pair.
Prune for provenance decay: every cited artifact must still resolve — the
file exists or the verdict digest is present under `.agents/ao/verdicts/`. A