Skills › Superpowers › diagnosing-superpowers › repeated-work.md
repeated-work.md
30 lines · skills/diagnosing-superpowers/prompts/repeated-work.md at 2ce8aa1
Most marks come from matching words in the file, so treat them as pointers.
1Read `prompts/analyst-common.md` first; it gives your role, inputs,
2context-safety rules, and the return format. This file adds the dimension.
4Dimension: Repeated work
6Find work the session did more than once.
81. Extract every tool call as `(line, turn, tool, key)` where `key` is: the
9 file path for reads/edits/writes; the command text for shell calls (strip
10 trailing whitespace; keep the whole command); the `description` plus the
11 first 80 characters of the prompt for subagent dispatches; the query for
12 searches.
132. Group by `(tool, key)` and report the groups at or over threshold:
15 | Category | Threshold | Exempt |
16 |---|---|---|
17 | reads, searches | 3 | |
18 | edits | 2 | |
19 | shell commands | 2 | status checks and test runs (`git status`, `ls`, `pwd`, test runners) |
20 | subagent dispatches | 2 with the same description | |
213. For each group, check whether anything changed between repetitions (a
22 write to that file, a compaction, a human correction). Say which case
23 it is; a re-read after an edit is not a finding, a re-read after a
24 compaction is a finding attributed to the compaction, a re-read with
25 nothing in between is a finding on its own.
264. Look for re-derived decisions: assistant text that reaches a conclusion
27 already stated earlier in the session (same file, same design choice,
28 same command to run). Quote both places.
295. One finding per group, with the first and last line numbers and the
30 count.
Read prompts/analyst-common.md first; it gives your role, inputs, context-safety rules, and the return format. This file adds the dimension.
Dimension: Repeated work
Find work the session did more than once.
- Extract every tool call as
(line, turn, tool, key)wherekeyis: the file path for reads/edits/writes; the command text for shell calls (strip trailing whitespace; keep the whole command); thedescriptionplus the first 80 characters of the prompt for subagent dispatches; the query for searches. - Group by
(tool, key)and report the groups at or over threshold:
| Category | Threshold | Exempt |
|---|---|---|
| reads, searches | 3 | |
| edits | 2 | |
| shell commands | 2 | status checks and test runs (git status, ls, pwd, test runners) |
| subagent dispatches | 2 with the same description |
- For each group, check whether anything changed between repetitions (a write to that file, a compaction, a human correction). Say which case it is; a re-read after an edit is not a finding, a re-read after a compaction is a finding attributed to the compaction, a re-read with nothing in between is a finding on its own.
- Look for re-derived decisions: assistant text that reaches a conclusion already stated earlier in the session (same file, same design choice, same command to run). Quote both places.
- One finding per group, with the first and last line numbers and the count.
Starts an agent subagent :11
Starts an agent dispatches :11