Reviewer tripwire (Calibration): a plan-mandated defect IS a finding,
reported as Important and labeled plan-mandated — the plan's authorship
does not grade its own work.
Controller rule (review loop): a plan-mandated finding, or any finding
conflicting with the plan's text, escalates to the human like any plan
contradiction — never dismissed because the plan mandates it.
E35 micro (frozen 0a98 replay, sonnet reviewer, 6v6): without the
tripwire 0/6 reports give the controller anything to escalate on (all
Approved, defect endorsed as spec-required); with it 6/6 report the
defect as a labeled finding.
Claims are fidelity and variance, not dollars (full attribution in the
superpowers-evals experiment log, 2026-06-11 L1 entry):
- Global Constraints header: 0/5 -> 5/5 adoption in micro-tests, exact
values verbatim; makes constraints mechanically propagatable to briefs
and reviewers (a version-floor violation class shipped because they
weren't). The one fix wave in the elicited full runs was a version-floor
catch this header enabled.
- Per-task Interfaces blocks: 0 -> 100% of tasks, exact signatures,
within-plan consistent; removes the controller's per-dispatch interface
re-derivation.
- Task right-sizing: 9.4 -> 8.4 mean tasks at svelte scale (kills
standalone Types/README micro-tasks); no effect at small scale.
- End-to-end (opus-written plan executed under SDD): guidance plan ran 1
fix wave vs control's 2-4 (control plan shipped a real Sierpinski bug);
execution cost equal within noise.
E30 replay: the planted-DRY catch is causally determined by the
controller-composed constraints block (0/6 with process-shaped vs 5/6
with the spec's own wording). E31 micro: this recipe doubles the rate
at which composed blocks carry the spec's cross-component relationship
(6/6 vs 3/6). Affects dev and the redesign equally (E29: both 4/5).
scripts/review-package generates the reviewer's input deterministically:
commit list, stat summary, and net diff with -U10 context, written to a
file from an explicit BASE. Live runs showed controllers improvising
'git diff HEAD~1..HEAD', which silently truncates multi-commit tasks,
and svelte's five fix dispatches shipped without re-running any tests —
fix dispatches now explicitly carry the implementer's
re-run-and-report contract.
The skill read as a changelog: 'combined task review,' 'one reviewer,
one reading,' 'one dispatch,' and an example still showing diffs pasted
into prompts. A reader who never saw the two-reviewer design has no
referent for 'combined.' Prose now states the design directly, and the
flowchart/example reflect the diff-file handoff.
Paste adoption stayed at 0/15 even as a Red Flag — and the controller's
reluctance is locally rational: pasting loads the diff into the (most
expensive) controller context permanently, while a reviewer self-fetch
costs a few cheap turns. The diff-file handoff is cheap for both sides:
the controller redirects git diff to /tmp without reading it, and the
reviewer gets the whole change in one Read call.
Fourth planted-defect failure mode: the implementer's self-report said
'noted mild structural duplication; left unabstracted per YAGNI' and the
reviewer deferred to that framing, rating the duplication no finding at
all. The pre-judging keeps relocating — controller prompt, then reviewer
calibration, now the implementer's report. Rationales are claims; they
never downgrade severity.
Adoption was 6/11 reviews on fractals and 0/17 on svelte when phrased
as guidance; reviewers without the diff re-derive it by hand, which is
the single largest remaining reviewer cost. Now a Red Flags Never entry
and a REQUIRED marker on the template placeholder.
With merged review, a planted verbatim-duplication defect shipped: the
reviewer rated it Minor (YAGNI) under the strict cannot-be-trusted
definition of Important, and the Minor-rolls-up rule meant no fix was
ever dispatched and the final review never saw the finding. Calibration
now names merge-blocking maintainability damage (verbatim duplication,
swallowed errors, assertion-free tests) as Important, and controllers
must paste accumulated Minor findings into the final review dispatch.
Iteration-1 profiling: implementers and per-dispatch overhead dominate
(429 of 686 subagent turns; controller coordination is half the dollars
and scales with dispatch count), reviewers are individually lean, and
the controller pasted the diff in only 2 of 22 review dispatches when
the guidance was phrased as optional.
Changes: spec-reviewer-prompt.md + code-quality-reviewer-prompt.md
replaced by task-reviewer-prompt.md (one reviewer, one reading of a
pasted diff, two verdicts: spec compliance ✅/❌/⚠️ and task quality);
one fix dispatch can address both kinds of findings; controller now
runs git diff itself and pastes it (imperative, not optional);
implementers run focused tests while iterating and the full suite once
before committing; flowchart, example, Red Flags, tool tables updated.
The broad final whole-branch review is unchanged.
Round-2 fractals eval regressed to 70min/32.2M tokens (vs round-1's
42.8min/14.5M) while reaching baseline-parity quality. Per-subagent turn
profiling attributed it to: haiku dispatches taking 2-3x the turns of
sonnet (678 of 1197 subagent turns), reviewers re-fetching diffs by hand
(518 Bash calls), and evidence-rule narration. Changes: turn-count-beats-
token-price model guidance; controllers paste small diffs into reviewer
prompts (reviewers then need few or no tool calls); evidence scoped to
findings and would-be-bare-yes checks; Important defined as cannot-trust-
until-fixed with coverage suggestions Minor; fixes dispatched only for
Critical/Important.
Resumed the offending eval controller session and asked it why it
pre-judged despite the rule being in context. Its retrospective: the
motive was avoiding a review loop, the abstract rule was read but not
applied at the moment it governs, and a phrase-level trigger ('do not
flag', 'at most Minor', 'don't treat X as a defect', 'the plan chose')
would have fired where the principle did not.
Second observed instance: with the Constructing Reviewer Prompts rule
already live, a controller still wrote 'do not treat that duplication as
a defect to fix — the plan chose it; you may note it as a Minor
observation at most' into a quality reviewer dispatch, fabricating plan
intent from the plan's example snippet. Promote the rule to the Red
Flags Never list and name the rationalization.
Live eval deliverables shipped five polish defects; tracing each through
the transcripts showed three mechanisms, each now addressed:
- reviewers answered pointed checklist items with unsupported yes
(evidence rule: every What-to-Check answer needs file:line evidence)
- no reviewer ever saw the design's global constraints (controllers now
paste binding constraints into task requirements)
- test output noise was invisible everywhere (pristine-output checks in
implementer self-review and quality review)
In live eval runs, controllers given judgment-based model selection
stopped passing a model at all; the omitted parameter inherits the
session's top-tier model, silently making every subagent maximally
expensive (one run dispatched 26/26 reviewers on the session model).
A live eval run of sdd-quality-reviewer-catches-planted-defect caught the
SDD controller fabricating a plan constraint and instructing the quality
reviewer not to flag the planted DRY violation. The duplication shipped.
Constructing Reviewer Prompts now bans suppression directives alongside
open-ended broadening directives.
Superpowers is a complete software development methodology for your coding agents, built on top of a set of composable skills and some initial instructions that make sure your agent uses them.
## We're Hiring!
We're hiring someone to help out full time with Superpowers community and code work.
You can read about the job at https://primeradiant.com/jobs/superpowers-community-engineer/
If this sounds like someone you know, definitely send them our way.
@@ -11,6 +11,9 @@ Execute plan by dispatching a fresh implementer subagent per task, a task review
**Core principle:** Fresh subagent per task + task review (spec + quality) + broad final review = high quality, fast iteration
**Narration:** between tool calls, narrate at most one short line — the
ledger and the tool results carry the record.
**Continuous execution:** Do not pause to check in with your human partner between tasks. Execute all tasks from the plan without stopping. The only reasons to stop are: BLOCKED status you cannot resolve, ambiguity that genuinely prevents progress, or all tasks complete. "Should I continue?" prompts and progress summaries waste their time — they asked you to execute the plan, so execute it.
## When to Use
@@ -79,6 +82,20 @@ digraph process {
}
```
## Pre-Flight Plan Review
Before dispatching Task 1, scan the plan once for conflicts:
- tasks that contradict each other or the plan's Global Constraints
- anything the plan explicitly mandates that the review rubric treats as a
defect (a test that asserts nothing, verbatim duplication of a logic block)
Present everything you find to your human partner as one batched question —
each finding beside the plan text that mandates it, asking which governs —
before execution begins, not one interrupt per discovery mid-plan. If the
scan is clean, proceed without comment. The review loop remains the net for
conflicts that only emerge from implementation.
## Model Selection
Use the least powerful model that can handle each role to conserve cost and increase speed.
@@ -88,6 +105,8 @@ Use the least powerful model that can handle each role to conserve cost and incr
**Integration and judgment tasks** (multi-file coordination, pattern matching, debugging): use a standard model.
**Architecture and design tasks**: use the most capable available model.
The final whole-branch review is one of these — dispatch it on the most
capable available model, not the session default.
**Review tasks**: choose the model with the same judgment, scaled to the
diff's size, complexity, and risk. A small mechanical diff does not need the
@@ -100,8 +119,10 @@ most expensive — which silently defeats this section.
**Turn count beats token price.** Wall-clock and context cost scale with how
many turns a subagent takes, and the cheapest models routinely take 2-3× the
turns on multi-step work — costing more overall. Use a mid-tier model as the
floor for implementers and reviewers; reserve the cheapest tier for
single-file mechanical fixes.
floor for reviewers and for implementers working from prose descriptions.
When the task's plan text contains the complete code to write, the
implementation is transcription plus testing: use the cheapest tier for
that implementer. Single-file mechanical fixes also take the cheapest tier.
"yes." A tight report that cites lines gives the controller everything
it needs.
Your final message is the report itself: begin directly with the
spec-compliance verdict. Every line is a verdict, a finding with
file:line, or a check you ran — no preamble, no process narration,
no closing summary.
## Calibration
Categorize issues by actual severity. Not everything is Critical.
@@ -123,6 +128,11 @@ Subagent (general-purpose):
would block a merge over — verbatim duplication of a logic block,
swallowed errors, tests that assert nothing. "Coverage could be broader"
and polish suggestions are Minor.
If the plan or brief explicitly mandates something this rubric calls a
defect (a test that asserts nothing, verbatim duplication of a logic
block), that IS a finding — report it as Important, labeled
plan-mandated. The plan's authorship does not grade its own work; the
human decides.
Acknowledge what was done well before listing issues — accurate praise
helps the implementer trust the rest of the feedback.
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
