great-cto 3.65.1 → 3.66.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "great-cto",
|
|
3
3
|
"description": "You already have the agent. This is everything around it. great_cto runs Claude Code as a pipeline of 72 specialist agents \u2014 an independent model checks each stage before the next builds on it, spending caps refuse rather than warn, and three decisions stay yours: what gets built, how, and whether it ships.",
|
|
4
|
-
"version": "3.
|
|
4
|
+
"version": "3.66.0",
|
|
5
5
|
"author": {
|
|
6
6
|
"name": "Alexander Velikiy",
|
|
7
7
|
"url": "https://hashnode.com/@Greatcto"
|
|
@@ -193,6 +193,8 @@ project: <basename of cwd>
|
|
|
193
193
|
confidence: medium|high # YOUR OWN rating. Never authoritative — see below
|
|
194
194
|
shape: A|B|C|D|E
|
|
195
195
|
eval: EVAL-<name>.md # optional — the eval whose agent this lesson changes
|
|
196
|
+
component: prompt|constraint|skill|tool|memory # what part of the harness it changes (ADR-038)
|
|
197
|
+
changes: skills/<name>/SKILL.md # only for component: skill — the skill it changes
|
|
196
198
|
---
|
|
197
199
|
|
|
198
200
|
**Two axes, and only one of them is earned.** `support:` is written by
|
|
@@ -249,7 +251,9 @@ a regression — measured by `node "$_LM" --measure <slug>` (prints the cost;
|
|
|
249
251
|
`--yes` spends it), which runs the lesson's `eval:` with and without the lesson.
|
|
250
252
|
A lesson with no `eval:` stays in its projects. Never write `eval_delta:` or
|
|
251
253
|
`eval_confidence:` yourself: `lessons-write.mjs` discards them, and a lesson
|
|
252
|
-
cannot vouch for itself.
|
|
254
|
+
cannot vouch for itself. Name the `component:` — a lesson without one cannot be
|
|
255
|
+
measured; a `skill` lesson is measured against a control arm (the hint alone), so
|
|
256
|
+
say which skill it changes. Do not run `--measure` or `--no-eval` on your own — they
|
|
253
257
|
spend money or bypass evidence, and that is the operator's call.
|
|
254
258
|
|
|
255
259
|
Flags: `--dry-run` previews without writing, `--force` re-promotes a slug already
|
package/package.json
CHANGED