Skip to content

armgit / hypit

Public Imported from GitHub
0

Clone any viral video with AI agents. Not just a script, the whole workflow: swap the face, the words, the B-roll, ship 100 variants in one command, and get your 100M views.

docs: give each observation to its own subagent where the harness has them

rponea518f8b07ba2
diff --git a/.agents/skills/hypit/references/reconstruction/index.md b/.agents/skills/hypit/references/reconstruction/index.md
index bff49905..39186066 100644
--- a/.agents/skills/hypit/references/reconstruction/index.md
+++ b/.agents/skills/hypit/references/reconstruction/index.md
@@ -92,7 +92,9 @@ the reference keeps it.
When the Vertex credentials are absent, `agent` is what can run — say that, say that it reads the
reference a little less closely, and take the author's answer. Then say which observer is reading,
-before the evidence starts. Both are in `observers.md`.
+before the evidence starts. Both are in `observers.md`, which also says to give each observation to
+its own subagent where the harness has them: that is what makes the sweep parallel rather than serial,
+and what keeps the observations independent of one another.
This is the only question this route asks about how it runs, and it is asked once, before the evidence
starts.
diff --git a/.agents/skills/hypit/references/reconstruction/observers.md b/.agents/skills/hypit/references/reconstruction/observers.md
index 5506e066..c45f5c5c 100644
--- a/.agents/skills/hypit/references/reconstruction/observers.md
+++ b/.agents/skills/hypit/references/reconstruction/observers.md
@@ -85,13 +85,35 @@ and do not put the gap to the author.
`voices` and the sound half of each `boundary` are the observations this shapes. They are attributions
read off pictures and word timings, and that is what a later reader should take them for.
-## The comparison in the loop is yours too
+## Answer each task in its own subagent when you can
-`compare_reconstruction` on the `agent` observer returns the two images and the same question rather
-than an answer. You look at the reference frame and your render and describe the differences yourself.
+Every task the tool returns is self-contained: a `key`, an `instruction`, a `prompt` that already
+carries the whole-reference context, and absolute `image_refs`. No task refers to another task's
+answer. So if the harness you are running in can spawn subagents — Claude Code and Codex can, and
+some cannot — give each task to its own, and run them together.
-That makes the report yours, and you know what you built, so the discipline that keeps it useful is
-explicit:
+Hand the subagent the `instruction`, the `prompt` verbatim and the `image_refs`, and nothing else.
+Not what you built, not which component drew anything, not what you expect it to find, and not another
+observation's answer. Take the returned text and write it with `record_observation` yourself, so one
+writer owns the cache.
+
+Two things follow from one task per subagent, and both of them are properties the `gemini` observer
+has for free:
+
+- **The observations are independent again.** Each one is answered by a reader that saw only its own
+ pictures and its own question, which is what a separate request to Gemini is. Two observations
+ agreeing is evidence again rather than one reader being consistent with itself.
+- **The comparison closes blind.** A subagent handed two unlabelled images and the comparison question
+ does not know which is the reference, what was built or what you hoped to see. That is the same
+ unlabelled pair the other observer gets.
+
+Answer the tasks in order yourself when the harness has no subagents. It is slower, and it is the
+fallback rather than the shape to aim for.
+
+## Comparing without subagents
+
+Answering the comparison yourself makes the report yours, and you know what you built, so the
+discipline that keeps it useful has to be explicit:
- **Write the differences down before naming a cause.** List what is visibly different, in the words
you would use if you had never seen the source. Deciding what went wrong first, then looking, finds
@@ -102,6 +124,9 @@ explicit:
is too wide — read the number off the frame against the frame's own width and height, and change the
value once. `reconstruction-loop.md` says why an attempt is for differences that have no number.
+The same three hold when a subagent compares, and they cost nothing there; what a subagent adds is
+that the comparer is not the builder.
+
## Running it
```text
diff --git a/.agents/skills/hypit/references/reconstruction/reconstruction-loop.md b/.agents/skills/hypit/references/reconstruction/reconstruction-loop.md
index bb914d11..318877de 100644
--- a/.agents/skills/hypit/references/reconstruction/reconstruction-loop.md
+++ b/.agents/skills/hypit/references/reconstruction/reconstruction-loop.md
@@ -62,11 +62,13 @@ very bias it claims to correct: it sees the reconstruction, knows what was built
expects. This is why a font that is wrong is caught — the comparison names it, and the package is
repaired, rather than a font being chosen because it resembles what you remember.
-**On the `agent` observer, that observer is you**, and the command returns the two images and the
-question instead of an answer. The bias above is then real and unavoidable, so `observers.md` states
-the discipline that replaces blindness: write the differences down before naming a cause, count only
-repairs aimed at a difference you wrote down, and read a quantity off the frame rather than spending
-an attempt guessing at it.
+**On the `agent` observer the command returns the two images and the question instead of an answer,
+and who looks at them is up to the harness.** Give them to a subagent when you can: one handed two
+unlabelled images and the question knows neither which is the reference nor what was built, which is
+the same blindness this section rests on. Answer it yourself only when there are no subagents, and
+then the bias above is real, so `observers.md` states the discipline that stands in for blindness —
+write the differences down before naming a cause, count only repairs aimed at a difference you wrote
+down, and read a quantity off the frame rather than spending an attempt guessing at it.
Hypit Studio is a browser preview for a person to look at. It is not a source of the image this
loop needs: that image is the local still render in `../preview.md`, which draws one element without