@sjawhar/pi-envoy 17.31.1 → 17.31.2
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.
- package/dist/envoy.js +1 -1
- package/dist/skills/agent-secrets/SKILL.md +3 -6
- package/dist/skills/approval-review/SKILL.md +8 -173
- package/dist/skills/approval-review/references/never-decide.md +1 -1
- package/dist/skills/envoy/SKILL.md +1 -6
- package/dist/skills/outside-message/SKILL.md +3 -8
- package/dist/skills/plain-words/SKILL.md +5 -7
- package/package.json +1 -1
- package/dist/skills/approval-review/references/reviewer-principles.md +0 -54
package/dist/envoy.js
CHANGED
|
@@ -47722,7 +47722,7 @@ var import_nats2 = __toESM(require_mod4(), 1);
|
|
|
47722
47722
|
|
|
47723
47723
|
// package.json
|
|
47724
47724
|
var name = "@sjawhar/pi-envoy";
|
|
47725
|
-
var version2 = "17.31.
|
|
47725
|
+
var version2 = "17.31.2";
|
|
47726
47726
|
var package_default = {
|
|
47727
47727
|
name,
|
|
47728
47728
|
version: version2,
|
|
@@ -62,8 +62,7 @@ agent-secrets LEGION_SLACK_BOT_TOKEN -- \
|
|
|
62
62
|
```
|
|
63
63
|
|
|
64
64
|
Your own shell expands `$LEGION_SLACK_BOT_TOKEN` before `agent-secrets` runs, and in your shell that
|
|
65
|
-
name is empty.
|
|
66
|
-
`{"ok":false,"error":"not_authed"}`. What the client hands over — the value, or the placeholder of
|
|
65
|
+
name is empty. Slack then answers `{"ok":false,"error":"not_authed"}`. What the client hands over — the value, or the placeholder of
|
|
67
66
|
a secret behind the airlock — exists only in the environment of the command the client starts, so
|
|
68
67
|
that command must read it itself. Put the command in single quotes, inside
|
|
69
68
|
`sh -c`, where the quotes keep the expansion for the child, and give `curl` the header on standard
|
|
@@ -236,8 +235,7 @@ A session on a person's own development box is not a pod, and `agent-secrets ide
|
|
|
236
235
|
for two reasons:
|
|
237
236
|
|
|
238
237
|
- **The machine is not logged in.** `agent-secrets machine login-status` prints `issued` only while
|
|
239
|
-
the machine holds its credential. The person logs the machine in
|
|
240
|
-
theorem-labs/agent-infra, "Log the VM in to the secrets broker", is that step.
|
|
238
|
+
the machine holds its credential. The person logs the machine in, as `/legion/broker/guides/log-a-machine-in/` in the documentation site shows.
|
|
241
239
|
- **The agent is not a broker session.** The helper knows only the process tree it registered. The
|
|
242
240
|
person restarts the agent as one:
|
|
243
241
|
`agent-secrets register --wait 10 --exec -- <the agent's command>`.
|
|
@@ -269,8 +267,7 @@ agent-secrets LEGION_SLACK_BOT_TOKEN --reason 'read the thread of a Slack mentio
|
|
|
269
267
|
--data-urlencode limit=200'
|
|
270
268
|
```
|
|
271
269
|
|
|
272
|
-
|
|
273
|
-
messages of the thread. A `ts` that names no message answers
|
|
270
|
+
It exits 0 and answers `"ok": true` with the messages of the thread. A `ts` that names no message answers
|
|
274
271
|
`{"ok":false,"error":"thread_not_found"}`, also at exit 0, which is why the body decides and not the
|
|
275
272
|
status. A thread read needs the app's `channels:history` scope in a public channel and
|
|
276
273
|
`groups:history` in a private one; without them Slack answers `missing_scope`, which only a person
|
|
@@ -8,13 +8,11 @@ description: "Use when you are asked to judge an approval request: a Dispatch re
|
|
|
8
8
|
An approval request names one or more documents at a version and asks whether that version may
|
|
9
9
|
stand. Your verdict is a prediction about a person: would the person whose approval requests this
|
|
10
10
|
deployment serves approve this version, or ask for changes? It is not a judgement of your own about
|
|
11
|
-
how good the document is. A document you would have written differently
|
|
12
|
-
|
|
11
|
+
how good the document is. A document you would have written differently may be one that person
|
|
12
|
+
approves. A document you find complete may be one they send back.
|
|
13
13
|
|
|
14
14
|
Two things follow. Judge the document against what that person decides, not against your taste. And
|
|
15
|
-
|
|
16
|
-
that measured this work found no recipe that beats approving everything, so nothing here lets you
|
|
17
|
-
decide an approval for anybody.
|
|
15
|
+
nothing here lets you decide an approval for anybody.
|
|
18
16
|
|
|
19
17
|
## What you never decide
|
|
20
18
|
|
|
@@ -50,149 +48,10 @@ is no rule. The reference's one question is what keeps the two costs apart.
|
|
|
50
48
|
where the words barely moved. Where the version you can read is not the version named, say so and
|
|
51
49
|
decide nothing.
|
|
52
50
|
|
|
53
|
-
## What was measured, and what it means
|
|
54
|
-
|
|
55
|
-
One round, on 2026-11-12: 27,540 judgements over 459 cases and 12 arms, each arm one model and one
|
|
56
|
-
information recipe, every arm over the same cases. The cases are drawn from that person's own
|
|
57
|
-
answered decisions, from a corpus pinned at event 396,268 and cut at 2026-11-05; the scored cases
|
|
58
|
-
were answered at or after that cut, so no recipe was built from them. The numbers below are the 250
|
|
59
|
-
approval cases of the report slice, five judgements per case, the modal verdict counted.
|
|
60
|
-
|
|
61
|
-
The baseline is approving everything: on those 250 cases that agrees with the person 0.792 of the
|
|
62
|
-
time (198 of 250). Each arm is paired against it over the same cases.
|
|
63
|
-
|
|
64
|
-
**Every interval in this file is two-sided at 90 per cent unless the line quoting it names another
|
|
65
|
-
level, and the level is said here, before the first table, because one run's verdict turns on
|
|
66
|
-
it.** The arms table below and the six runs further down state exact McNemar intervals, one arm or
|
|
67
|
-
one run each; the pooled figures state bootstrap intervals. The three upper bounds under "No arm
|
|
68
|
-
reached the rate the person set" are one-sided, and say so where they are quoted.
|
|
69
|
-
|
|
70
|
-
| arm | agreement | against the baseline |
|
|
71
|
-
| --- | --- | --- |
|
|
72
|
-
| many-shot-fable-5-1 | 0.780 (195 of 250) | −0.012 in [−0.037, +0.017], McNemar p 0.607 |
|
|
73
|
-
| many-shot-opus-5-5 | 0.772 (193) | −0.020 in [−0.049, +0.014], p 0.383 |
|
|
74
|
-
| both-opus-5-5 | 0.736 (184) | −0.056 in [−0.095, −0.011], p 0.038 |
|
|
75
|
-
| bare-opus-5-5 | 0.656 (164) | −0.136 in [−0.165, −0.093], p 0.000 |
|
|
76
|
-
| distilled-opus-5-5 | 0.492 (123) | −0.300 in [−0.362, −0.228], p 0.000 |
|
|
77
|
-
| distilled-jev | 0.280 (70) | −0.512 in [−0.592, −0.422], p 0.000 |
|
|
78
|
-
|
|
79
|
-
Read in order, that says four things.
|
|
80
|
-
|
|
81
|
-
**No arm beat approving everything.** The two best are indistinguishable from it; every other arm is
|
|
82
|
-
worse than it, and most by a wide margin.
|
|
83
|
-
|
|
84
|
-
**No arm reached the rate the person set.** The rate is one disagreement in twenty. No arm of the
|
|
85
|
-
twelve fitted a threshold that holds it: the lowest bound any confidence band of any arm reaches is
|
|
86
|
-
0.279, and coverage at that rate is zero for all twelve. Nor does any shortcut reach it. No stratum
|
|
87
|
-
of the person's own decisions holds a request-changes rate under 5 percent — the lowest bound is
|
|
88
|
-
0.130, over 8.6 percent of the volume — and no consensus across model families reaches it either:
|
|
89
|
-
the best is many-shot with all three models agreeing, coverage 0.824 at a bound of 0.229, where
|
|
90
|
-
approving everything on those same cases disagrees at 0.184. Those three bounds — 0.279, 0.130
|
|
91
|
-
and 0.229 — are one-sided Clopper-Pearson upper bounds at 90 per cent, not two-sided intervals:
|
|
92
|
-
each says the rate is at most that number and says nothing about how low it is.
|
|
93
|
-
|
|
94
|
-
**Retrieval is the one thing that helped, and the measurement that says so is of stateless calls
|
|
95
|
-
rather than of an agent.** The many-shot recipe, which puts the nearest past decisions in front of
|
|
96
|
-
the model, is worth 0.116 more agreement than the bare recipe on the same model: 0.772 against
|
|
97
|
-
0.656. Both of those arms are one model call with no tools, so what the number supports is the
|
|
98
|
-
information and not a way of working. The step below is how you get it.
|
|
99
|
-
|
|
100
|
-
**The written distillation made agreement worse, and the error it adds has one direction.** The
|
|
101
|
-
errors of the two distilled arms are lopsided: `distilled-opus-5-5` made 105 false requests for
|
|
102
|
-
changes against 22 false approvals, and `distilled-jev` 177 against 3. A written distillation of
|
|
103
|
-
what makes a person ask for changes reads, to a model given it as instructions, as a list of reasons
|
|
104
|
-
to ask for changes, so the model asks for changes.
|
|
105
|
-
|
|
106
|
-
So this skill does not tell you that you may decide an approval for anyone. It tells you what to
|
|
107
|
-
read, what never to decide, and what the measurement supports. Where a deployment wants verdicts
|
|
108
|
-
decided without a person, the number it must beat is 0.792 on these cases, and nothing measured here
|
|
109
|
-
beats it.
|
|
110
|
-
|
|
111
|
-
One rule was validated: the never-decide rule above held 60 of 60 cases of a set written blind and
|
|
112
|
-
met once, against 15 of 60 for the pattern-matching version it replaced, and it decided every
|
|
113
|
-
ordinary document rather than holding them too.
|
|
114
|
-
|
|
115
|
-
**What this text is worth depends on whether the agent reading it can retrieve, and one run of
|
|
116
|
-
that comparison cannot tell you.** A fresh agent was given this skill and nothing else and judged
|
|
117
|
-
250 of the person's approval decisions; the same agent with no skill at all judged the same 250,
|
|
118
|
-
on `claude-opus-5-5`, one judgement per case. The comparison was run **six times**, nothing
|
|
119
|
-
changed between the runs of one build, and this is what the six printed:
|
|
120
|
-
|
|
121
|
-
| run | skill arm | control | paired difference | what the program called it | the skill it read |
|
|
122
|
-
| --- | --- | --- | --- | --- | --- |
|
|
123
|
-
| 1 | 0.668 (167) | 0.588 (147) | +0.080 in [+0.040, +0.109] | passes | `586bb99f` |
|
|
124
|
-
| 2 | 0.644 (161) | 0.620 (155) | +0.024 in [−0.018, +0.063] | fails | `19029752` |
|
|
125
|
-
| 3 | 0.688 (172) | 0.640 (160) | +0.048 in [+0.007, +0.082] | passes | `19029752` |
|
|
126
|
-
| 4 | 0.700 (175) | 0.636 (159) | +0.064 in [+0.019, +0.101] | passes | `19029752` |
|
|
127
|
-
| 5 | 0.652 (163) | 0.644 (161) | +0.008 in [−0.034, +0.048] | fails | `19029752` |
|
|
128
|
-
| 6 | 0.676 (169) | 0.628 (157) | +0.048 in [+0.008, +0.080] | passes | `abcffa20` |
|
|
129
|
-
|
|
130
|
-
**Four passes and two failures over the same 250 cases, the same two arms and the same model.**
|
|
131
|
-
The six runs read three builds of this text, which differ in this section and nowhere else, and
|
|
132
|
-
the four runs of one build, `19029752`, are themselves two passes and two failures. The runs
|
|
133
|
-
differ by 0.072, which is about the width of one run's own interval, so no single run of this
|
|
134
|
-
comparison carries a number worth recording. Both arms move together, because the noise is the
|
|
135
|
-
model's own run and is shared by them; what survives it is the paired difference, and it survives
|
|
136
|
-
only when each case is averaged over the runs.
|
|
137
|
-
|
|
138
|
-
**So the number this skill records is the pooled one: +0.036 in [+0.010, +0.063], over the four
|
|
139
|
-
runs of build `19029752`, the only build of this text more than one run read.** Each of the 250
|
|
140
|
-
cases carries the mean of its outcome over those four runs, the estimate is the mean of the 250
|
|
141
|
-
case-level differences, and the interval is a paired bootstrap of 20,000 resamples of those
|
|
142
|
-
differences under seed 7, **two-sided at 90 per cent**. The bootstrap holds the runs fixed, so it
|
|
143
|
-
carries none of the 0.056 those four runs spread over. Quote it with that spread or not at all.
|
|
144
|
-
|
|
145
|
-
**The number this skill records is the same at either level**: read at 95 per cent off the same
|
|
146
|
-
draw it is +0.036 in [+0.005, +0.068], which still excludes zero. Of the six runs, two hold zero
|
|
147
|
-
at both levels — runs 2 and 5, the two the program already called failures at 90 — and
|
|
148
|
-
**run 3 holds zero at 95 per cent and not at 90**: its interval is [+0.007, +0.082] at 90 and
|
|
149
|
-
[−0.00002, +0.087] at 95, so at the stricter level the table would read three passes and three
|
|
150
|
-
failures. Read that margin before you quote it: run 3 does not clear zero at 95 per cent by any
|
|
151
|
-
distance a reader should weigh, it touches it, and "three and three" is a fact about the fourth
|
|
152
|
-
decimal of one run. Which is the argument for the sentence above rather than against it — no
|
|
153
|
-
single run's verdict is the finding, the recorded number is the pooled one, and it does not move.
|
|
154
|
-
A paired bootstrap over the same six runs agrees with these intervals on every row but that one,
|
|
155
|
-
which is the row sitting on the line rather than two instruments that disagree.
|
|
156
|
-
`pool-runs.ts` prints both levels at every reading, drawn from one set of resamples under one
|
|
157
|
-
seed, so the two differ by the level and by nothing else.
|
|
158
|
-
|
|
159
|
-
**The other three numbers in the table, and what each of them is.** `586bb99f` and `abcffa20`
|
|
160
|
-
were read by one run each, +0.080 in [+0.040, +0.109] and +0.048 in [+0.008, +0.080]. Those are
|
|
161
|
-
single runs' own numbers, which the paragraph above says are not worth recording, and they are
|
|
162
|
-
here because they are all those two builds have. Over all six runs the reading is +0.045 in
|
|
163
|
-
[+0.022, +0.070], which is a number of no build at all, since a number pooled over runs of three
|
|
164
|
-
builds is a number of none of them; it is worth having as the wider evidence that the three agree,
|
|
165
|
-
and it is not a number to quote on its own.
|
|
166
|
-
|
|
167
|
-
**The last column is the build each run read, and no run of this comparison can ever be of the
|
|
168
|
-
bytes you are reading**, because writing a measured number into a file changes the file. That is
|
|
169
|
-
why the figures are read per build at all. The three builds differ in this section and nowhere
|
|
170
|
-
else. Leaving out the two runs of the other builds lowers the number from +0.045 to +0.036, which
|
|
171
|
-
is the direction that costs this text rather than flatters it.
|
|
172
|
-
`scripts/approval-eval/pool-runs.ts` reads each run's own copy of this file, groups the runs by
|
|
173
|
-
its hash and prints all of that, so nothing above is a hash or a direction anybody typed.
|
|
174
|
-
|
|
175
|
-
**And the second shape, which measures something else.** With the nearest past decisions placed in
|
|
176
|
-
a file beside each case and both arms told of it, both reached 0.784 (196 of 250), a difference of
|
|
177
|
-
+0.000 in [−0.024, +0.024]. The agent without this skill read that file unprompted in every case
|
|
178
|
-
it was watched on, so both arms performed the retrieval recipe: that comparison is of this text on
|
|
179
|
-
top of behaviour the model already has.
|
|
180
|
-
|
|
181
|
-
Neither shape measures the 0.116 above: the first gives the recipe to neither arm and the second
|
|
182
|
-
to both, and no agent on that machine could retrieve live. What the pair says is that this text is
|
|
183
|
-
worth a few points of agreement to an agent that cannot retrieve, and nothing measurable to one
|
|
184
|
-
already doing what the text describes.
|
|
185
|
-
|
|
186
|
-
Read the numbers together before you quote any of them. The never-decide rule has its own evidence
|
|
187
|
-
and it is strong. The retrieval recipe's 0.116 is real and is of stateless calls. This text's own
|
|
188
|
-
gain against nothing is +0.036 in [+0.010, +0.063], over the four runs of the one build of it that
|
|
189
|
-
more than one run read, and those four runs spread over 0.056; it is zero against an agent already
|
|
190
|
-
retrieving.
|
|
191
|
-
|
|
192
51
|
## The retrieval step
|
|
193
52
|
|
|
194
|
-
This is the many-shot
|
|
195
|
-
|
|
53
|
+
This is the many-shot method: it puts the nearest past decisions of the person in front of you, with
|
|
54
|
+
the `dispatch` command and nothing else.
|
|
196
55
|
|
|
197
56
|
1. Write the case in a sentence or two of its own words: what the document asks for, in the terms it
|
|
198
57
|
uses. That is your query.
|
|
@@ -206,8 +65,8 @@ part the measurement supports.
|
|
|
206
65
|
4. Judge the case with those decisions in front of you, and say how many you found. A case with two
|
|
207
66
|
neighbours is not a case with eight.
|
|
208
67
|
|
|
209
|
-
Two filters
|
|
210
|
-
|
|
68
|
+
Two filters keep the method honest, and dropping either turns it into a way of reading the answer
|
|
69
|
+
off the board:
|
|
211
70
|
|
|
212
71
|
- **Only decisions the person answered before the case you are judging.** A later decision is
|
|
213
72
|
information you would not have had, and a method that uses one cannot be run live.
|
|
@@ -216,26 +75,10 @@ answer off the board:
|
|
|
216
75
|
That is the label, not a neighbour. Drop every hit on the judged document or issue, however near,
|
|
217
76
|
and count how many you dropped.
|
|
218
77
|
|
|
219
|
-
In the round, retrieval dropped 1,920 neighbours for sitting on the judged case's own ground, and no
|
|
220
|
-
scored row carried a neighbour naming the judged case's own document at any version.
|
|
221
|
-
|
|
222
|
-
## The distillation
|
|
223
|
-
|
|
224
|
-
`skill://approval-review/references/reviewer-principles.md` is the written distillation: what makes
|
|
225
|
-
that person approve, what makes them ask for changes, what they always want said, and the tone they
|
|
226
|
-
refuse. It quotes the person and keeps their wording, so leave its text as it stands rather than
|
|
227
|
-
correcting its words to this repository's house style.
|
|
228
|
-
|
|
229
|
-
Read it as evidence about what that person decides on, which is what it is good for, and which is
|
|
230
|
-
useful to a person learning the same ground. Do not treat it as a checklist to run a document
|
|
231
|
-
against. Given to a model as its instructions it made agreement fall — 0.492 against the bare
|
|
232
|
-
recipe's 0.656 on the same model, and 0.280 on the model that took it hardest — and it made the
|
|
233
|
-
model ask for changes far more often than the person does. The numbers are in the section above.
|
|
234
|
-
|
|
235
78
|
## The share that still reaches the person
|
|
236
79
|
|
|
237
80
|
If a deployment ever turns deciding on, a share of what the agent would decide goes to the person
|
|
238
|
-
anyway. That share is
|
|
81
|
+
anyway. That share is a measurement that never stops, so it has to look like every other request:
|
|
239
82
|
|
|
240
83
|
- The sampled requests are unmarked. Nothing in a request, its wording, its order or its timing says
|
|
241
84
|
that this one was sampled.
|
|
@@ -246,11 +89,3 @@ anyway. That share is the measurement that never stops, so it has to look like e
|
|
|
246
89
|
|
|
247
90
|
Until a deployment turns deciding on, every request goes to the person and the agent's verdict is a
|
|
248
91
|
record of what it would have answered.
|
|
249
|
-
|
|
250
|
-
## What this skill does not tell you
|
|
251
|
-
|
|
252
|
-
That agreement measured on past decisions holds on next week's documents. Everything above is read
|
|
253
|
-
off decisions the person had already made, over one corpus and one cut. A person's view moves, the
|
|
254
|
-
documents put in front of them change, and the one thing no record holds is the spec nobody wrote
|
|
255
|
-
because its author expected a refusal. Treat every number here as a reading of that round, and
|
|
256
|
-
nothing more.
|
|
@@ -11,7 +11,7 @@ under and you hand it back to the person.
|
|
|
11
11
|
|
|
12
12
|
**Does the document ask for the act, or does it only mention it?**
|
|
13
13
|
|
|
14
|
-
A request
|
|
14
|
+
A request is often a technical document, and a technical document is full of deletions,
|
|
15
15
|
credentials, vendors and organization settings that it only describes. Describing is not asking.
|
|
16
16
|
Ask yourself what follows from a person writing "approved" under this document as it stands:
|
|
17
17
|
|
|
@@ -144,12 +144,7 @@ Every topic starts with `notifications.`.
|
|
|
144
144
|
Neither a kept answer nor an error releases the role's claim: a holder whose `holder_state` is `unresponsive` keeps its role.
|
|
145
145
|
`envoy_role_get(role=...)` finds the holder first.
|
|
146
146
|
The forward is kept on the holder's inbox, so the holder's session gets it once from its own durable consumer, also after a restart, a reconnect, or while it was not running.
|
|
147
|
-
From a shell with no tool device,
|
|
148
|
-
|
|
149
|
-
- A message that must reach more than 4 role topics is the project controller's to send, not yours.
|
|
150
|
-
Each publish wakes a session, and a woken session reads its whole context again, which the provider charges as uncached input: a mean of 576,083 tokens a wake, a median of 493,636 and a ninetieth percentile of 1,327,888.
|
|
151
|
-
The controller is also the one session that can read whether a wave of wakes is already running, and a fan-out sent into a daemon restart or a pod roll is what puts an Anthropic organization over its per-minute uncached-input cap.
|
|
152
|
-
Send the controller the topics and the text, and it sends them at its own rate (`skill://legion-controller`).
|
|
147
|
+
From a shell with no tool device, such as one in a Legion pod, read the holder with `curl -sS -H @<(printf 'Authorization: Bearer %s' "$(< "$ENVOY_TOKEN_FILE")") "$ENVOY_URL/v1/roles/<role>" | jq -r '.holder // .reason // .error'` in bash (the `<(...)` keeps the token out of `curl`'s arguments), which prints the holder's session id, `unclaimed` for a role nobody holds, or the error of a refused read, such as `unauthorized` for a bad token.
|
|
153
148
|
|
|
154
149
|
- For a team, the group subscribes to `notifications.team.<name>.>`, and any member publishes to `notifications.team.<name>.all`.
|
|
155
150
|
Envoy's stream keeps a team topic, so a publish returns once it is stored and every session that follows the topic gets it once from its own consumer
|
|
@@ -6,8 +6,6 @@ description: "Use before any text reaches a person outside the company: a commen
|
|
|
6
6
|
# A message to someone outside the company
|
|
7
7
|
|
|
8
8
|
The operator approves every message that leaves the company, and four steps come before you ask.
|
|
9
|
-
The rules below are the operator's own, from the answers that set them; this skill is where they
|
|
10
|
-
are written down.
|
|
11
9
|
|
|
12
10
|
## The exclusions
|
|
13
11
|
|
|
@@ -19,8 +17,7 @@ answer names such a project, and your stream's brief is where that answer reache
|
|
|
19
17
|
answer for the project in front of you, there is no exclusion.
|
|
20
18
|
|
|
21
19
|
**A pull request to a project that has already enrolled**, with its description and its comments to
|
|
22
|
-
that project's maintainers.
|
|
23
|
-
enrolled project, and ceremony around a pull request must not stop the work. The four steps still
|
|
20
|
+
that project's maintainers. The four steps still
|
|
24
21
|
apply. The approval ask does not. The work does not wait for the pull request either: prepare and
|
|
25
22
|
sign the changes locally, and send them when the work they belong to ends.
|
|
26
23
|
|
|
@@ -54,8 +51,7 @@ reputation for low-quality or careless submissions.
|
|
|
54
51
|
do: a reader of the same model as the writer is the weaker check, though no measurement here
|
|
55
52
|
separates the two.
|
|
56
53
|
|
|
57
|
-
**Step 4 ends with a list of what it removed, and that list goes in the ask.**
|
|
58
|
-
does not make the cut happen, so the list is what proves it did: three lines naming what came out,
|
|
54
|
+
**Step 4 ends with a list of what it removed, and that list goes in the ask.** The list is what proves the cut happened: three lines naming what came out,
|
|
59
55
|
or the sentence "the reviewer removed nothing", which is itself an answer.
|
|
60
56
|
|
|
61
57
|
**The text the operator approves is the text the reviewer cut, byte for byte.** Any edit after the
|
|
@@ -74,8 +70,7 @@ cover how the text reads.
|
|
|
74
70
|
## Then the ask
|
|
75
71
|
|
|
76
72
|
**Put the full message text in the ask itself.** Not a link to a document holding it, not a summary
|
|
77
|
-
of it, not "the draft is at …".
|
|
78
|
-
almost every one of these asks, and cannot judge the message from the ask.
|
|
73
|
+
of it, not "the draft is at …". A link makes the operator open another document, and the operator cannot judge the message from the ask.
|
|
79
74
|
|
|
80
75
|
An ask for an outside message has, in this order:
|
|
81
76
|
|
|
@@ -5,8 +5,7 @@ description: "Use before you send or publish any text a person reads: an ask, a
|
|
|
5
5
|
|
|
6
6
|
# Say the literal thing
|
|
7
7
|
|
|
8
|
-
|
|
9
|
-
thing instead.
|
|
8
|
+
Delete the empty verb phrases agents keep using, and say the literal thing instead.
|
|
10
9
|
|
|
11
10
|
Run this over your text before you send it. It takes a minute and it is not optional.
|
|
12
11
|
|
|
@@ -23,8 +22,7 @@ Run this over your text before you send it. It takes a minute and it is not opti
|
|
|
23
22
|
|
|
24
23
|
A fresh session that has only what the reader has removes what the reader does not need, and lists
|
|
25
24
|
every term it cannot define. Fix each listed term. A reviewer that reads only to approve has not
|
|
26
|
-
done this step,
|
|
27
|
-
ran. The check runs on the text going to the reader, never on the instructions you give that
|
|
25
|
+
done this step, so the two lists are what show it ran. The check runs on the text going to the reader, never on the instructions you give that
|
|
28
26
|
session or on what it sends back, which would never end.
|
|
29
27
|
|
|
30
28
|
## The test
|
|
@@ -32,7 +30,7 @@ session or on what it sends back, which would never end.
|
|
|
32
30
|
Take the verb phrase out and ask: **what literally happens?** Write that. If you cannot say what
|
|
33
31
|
happens, you did not know, and the phrase hid it from you as well as from the reader.
|
|
34
32
|
|
|
35
|
-
## The phrases
|
|
33
|
+
## The phrases
|
|
36
34
|
|
|
37
35
|
| Never | What to write |
|
|
38
36
|
|---|---|
|
|
@@ -43,12 +41,12 @@ happens, you did not know, and the phrase hid it from you as well as from the re
|
|
|
43
41
|
| carry it to X, carry the finding | tell X; send it to X; put it on the issue |
|
|
44
42
|
| this shows the gap, the number shows | the gap is 12 of 30; the number is 1.17 |
|
|
45
43
|
|
|
46
|
-
|
|
44
|
+
You reach for "land"
|
|
47
45
|
when you cannot say whether the thing merged, released or deployed. If you can say which, say it.
|
|
48
46
|
If you cannot, that is the sentence — "it is approved and not merged" — and it is more useful than
|
|
49
47
|
any verb.
|
|
50
48
|
|
|
51
|
-
On the "name the X" row
|
|
49
|
+
On the "name the X" row, use the test: delete "name". If the
|
|
52
50
|
sentence then says what happens, keep it that way; if it says nothing, you had not decided the
|
|
53
51
|
action, so decide it and write it.
|
|
54
52
|
|
package/package.json
CHANGED
|
@@ -1,54 +0,0 @@
|
|
|
1
|
-
This file tells a judging agent how the reviewer decides, and it is drawn from his decisions made before 2026-11-05.
|
|
2
|
-
|
|
3
|
-
## What makes him approve
|
|
4
|
-
|
|
5
|
-
1. **Standard practice.** He approves designs that use the documented, widely used solution for the problem, and he rejects hand-rolled scripts, proxies and tunnels that do the same job. To a "which option" ask he answers "do the standard thing," and any deviation needs a stated reason.
|
|
6
|
-
2. **Root-cause fix at the right layer.** He approves changes that remove a guard, workaround, hack or polling sleep and fix the cause, even when the hack works today. Fork changes must go upstream, become an extension, or be dropped, never held in the fork.
|
|
7
|
-
3. **Obvious or already-settled fixes.** He approves a fix to a plain bug or a task he already assigned. He also complains that it reached him at all.
|
|
8
|
-
4. **Parallel work.** He approves running independent lines of work at once across the available agents. He removes any ordering, waves or agent caps that are not true dependencies.
|
|
9
|
-
5. **Staged and measured steps.** He approves doing the safe part now, measuring it, then extending, and he approves gathering real data to settle a choice that would otherwise be argued. A prototype or time-boxed test passes if it has a defined metric and takes nothing from committed work.
|
|
10
|
-
6. **Rejected part gone, scope narrowed.** He approves once the revision deletes what he objected to and narrows to what he asked for. A pure reversion should be framed as only that.
|
|
11
|
-
7. **Cleanup after reading.** He approves deleting unused resources, branches, machines and stale issues. Someone must first read the items and keep what is still useful.
|
|
12
|
-
8. **Better than the baseline.** He accepts output that differs from upstream behaviour when the difference is a fix. He does not demand bug-for-bug copies.
|
|
13
|
-
9. **Delegation to an independent review.** He approves a long or urgent spec without reading it if an independent adversarial red-team will check it. The author may then apply agreed changes without coming back.
|
|
14
|
-
|
|
15
|
-
## What makes him ask for changes
|
|
16
|
-
|
|
17
|
-
10. **Asks within the agent's authority.** He rejects asks for logins, credential grants, permissions, merges, restarts, approvals, disk space or reply drafting, and points to the broker or an agent on the dev box ("I am not the gate"). He acts at once, tersely, only on what a person alone can do, such as a console click, a login code or an app install.
|
|
18
|
-
11. **Invented rules or blockers.** He orders deleted any rule, limit, ban, embargo, approval gate or access restriction he never set, and any text that makes his words stricter. The agent must check its premise first.
|
|
19
|
-
12. **Unasked code, checks and work.** He rejects new merge checks, string-matching guards, scripts, gates, speculative fixes and absolute bans written into code that he did not ask for. For a behaviour problem, an instruction or short skill must be tried and shown to fail first, built so it is cheap to delete.
|
|
20
|
-
13. **Repeat asks.** He rejects re-asking an answered question, asking about something he already did, or asking to re-approve corrected text. He expects his earlier general criteria applied to new cases.
|
|
21
|
-
14. **Deferral and waiting.** He rejects "out of scope," follow-up issues that hide unfinished work, parked or half-built parts, and pauses until others act. Every supported runtime gets the same behaviour and proof now, work continues locally, and deferral is fine only when he offers it.
|
|
22
|
-
15. **Long text.** Asks, specs, descriptions and outside messages over a paragraph come back as too long. He wants one sentence or bullets, with reasoning in a linked document.
|
|
23
|
-
16. **Fragmented asks and decisions pushed to him.** He rejects asks split across issues or documents, duplicate issues, per-item choice lists, and requests that he pick thresholds or names the data should settle. He wants one plan in one place, built with peers, after looking at data and transcripts first.
|
|
24
|
-
17. **Weak method or shortcuts on integrity.** He rejects evals and experiments that cannot prove or refute the claim, such as those with missing controls, confounds, leaked answers, an easier proxy question, or only easy cases. He also rejects graders that credit work failing the spec, redefined goals, and caches that let stale output pass.
|
|
25
|
-
18. **Small or oversized targets.** He rejects easy trims while the largest waste remains, and modest goals when a far better one is obvious. He also rejects proof schemes far larger than needed, calling both sandbagging or waste.
|
|
26
|
-
19. **Wrong target, priority or audience.** He says drop or close work he judges unimportant now, and he redirects work aimed at the wrong audience or at other people's machines and habits when the problem is "ours." Designs decided without discussing them with him go back to the start.
|
|
27
|
-
20. **Unready or unchecked requests.** He rejects requests sent before revisions, own checks and every comment are done, and designs not checked against real users' workflows and delivered artifacts. A UI design needs a mock-up or a running stub he can try.
|
|
28
|
-
21. **Judgement in the orchestrator.** He rejects a central daemon that decides merges, reads CI, guards rules or routes requests. It should be a plain state machine that relays events, with judgement left to agents and policy left to repository settings.
|
|
29
|
-
|
|
30
|
-
## What he always wants said
|
|
31
|
-
|
|
32
|
-
22. **Outside text that passed the full process.** Outside text must show research into the audience's conventions, templates, history and style. It must also pass simplified English, anti-slop, cross-model red-team and fresh-reader passes, contain only tested code, and carry no codenames, internal process, empty verbs, unfinished-sounding wording or present-versus-future caveats.
|
|
33
|
-
23. **Measured numbers tied to the top metric.** He wants real figures, with unmeasured costs measured rather than flagged and speedups stated against total runtime. Each figure should show its effect on throughput of real output at controlled cost, and numbers the reader cannot use, such as prices, IDs and hashes, should be struck.
|
|
34
|
-
24. **What the problem is and why.** When he cannot follow a project he wants an explainer: what existed, what changed, the target, the distance left, and why act. The root cause must come from transcripts and logs by asking further whys, not from a restated symptom.
|
|
35
|
-
25. **Full eval write-ups, surfaced to him.** A report needs the question, design and controls, evidence for and against, a test that could refute each claim, graphs, code and data, red-team results, and a home people will find. He does not count a deliverable as done until it is shown to him.
|
|
36
|
-
26. **Configurable, modular parts.** He wants per-project settings where the right value is uncertain, and optional components not every deployment must carry. Deployment-specific logic is added separately, not built into a general tool's core.
|
|
37
|
-
27. **A recovery path.** He wants the least disruptive recovery tried first, a report only after it fails, and a break-glass path for emergency changes.
|
|
38
|
-
|
|
39
|
-
## What he never accepts
|
|
40
|
-
|
|
41
|
-
28. **A rejected option returning.** He refuses any option he already rejected, even when it is reworded.
|
|
42
|
-
29. **Personal, ad hoc or duplicated infrastructure.** He refuses personal logins, keys or machines used as workarounds, as well as hand-minted credentials, standalone tunnels, cron jobs or units on the dev box, and new repositories for small tools. He wants service accounts or federated identity with no shared passwords, one source of truth, and nothing public that could stay private.
|
|
43
|
-
30. **Restricted team access.** He refuses guards that keep engineers from shared data or secrets they should have, data pools split by model, and tools limited to his machines. Only root or seed credentials and new applications need his approval.
|
|
44
|
-
|
|
45
|
-
## How he writes and what tone he refuses
|
|
46
|
-
|
|
47
|
-
- He writes short, blunt, often profane replies, often a single imperative such as "Drop this," "So fix it" or "Close this."
|
|
48
|
-
- He also dictates long run-on passages that end in a principle, and he expects it applied everywhere, not turned into a narrow rule.
|
|
49
|
-
- He quotes fragments of the text to aim at the exact sentence, and asks rhetorical questions such as "Why did this come to me?"
|
|
50
|
-
- He thanks only for real progress or a useful sanity check, and he says plainly when he was wrong.
|
|
51
|
-
- He refuses on tone alone any approval-seeking, hedged or soothing document with long caveats, option menus or repeated asks.
|
|
52
|
-
- He refuses dense, padded, agent-styled prose that buries the decision, coins terms he cannot parse, or reads like private notes.
|
|
53
|
-
- He refuses crisis framing of a non-issue, invented risks, and defensive excuses about scope, blockers or dependencies.
|
|
54
|
-
- He accepts short documents that open with the decision, give numbers, and state what is unproven.
|