ed/manual_slop
Private
Public Access
0
0
Fork 0
Code Activity

conductor(track): superpowers review nagent_takeaways_superpowers_20260619.md filled (5-part bridge)

Browse Source
This commit is contained in:
Edward R. Gonzalez
2026-07-05 13:38:27 -04:00
parent 1e4e31cc52
commit 049774d620
1 changed files with 29 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files
conductor/tracks/superpowers_review_20260619/nagent_takeaways_superpowers_20260619.md
+29 -5
View File
@@ -6,20 +6,44 @@ This bridge connects the superpowers review's verdicts to nagent_review's 16 fut
## TL;DR
[1 paragraph: written in Phase 7]
The superpowers review's 19 verdicts (14 superpowers-plugin skills + 5 MMA skills) overlap with nagent_review's 14-section analysis on **durable work** + **disposable workers** + **explicit interfaces** + **editable state** + **controlled writes**, but they diverge on the **meta-tooling vs application boundary** and on **TDD-with-pressure-scenarios for skills**. The superpowers review adds 3 new candidates not in nagent_review (see below). The two corpora reinforce each other on the "project follows the discipline" pattern but disagree on the dual-convention question (`docs/superpowers/specs/` vs `conductor/tracks/<id>/spec.md`); this is a user-driven choice, not a contradiction. The superpowers review recommends 25 deferred-rebuild changes (3 HIGH, 5 MEDIUM, 17 LOW); nagent_review recommends ~10 follow-up tracks. Together they form the input set for the user's deferred agent-directive rebuild.
## Cross-Reference Table
[10-15 rows: written in Phase 7]
| Superpowers section | Verdict | Nagent candidate | Relationship |
|---|---|---|---|
| §2 brainstorming | PARTIAL + INTEGRATE-PARTIAL | §3.6 ("instruction priority" — person-level discipline vs project-level) | Independent |
| §4 test-driven-development | PARITY + INTEGRATED | §10 ("durable tests" — nagent's principle that tests are durable artifacts) | Subsumes |
| §5 verification-before-completion | PARITY + INTEGRATED | §10 (same as above) | Subsumes |
| §6 systematic-debugging | PARITY + INTEGRATED | §10 (anti-rationalization as evidence-based discipline) | Extends |
| §7 subagent-driven-development | PARITY + INTEGRATED | §5 (durable-work thesis shapes subagent dispatch) | Subsumes |
| §9 dispatching-parallel-agents | PARTIAL + INTEGRATED | §5 (project's "disposable workers" thesis) | Independent (project chose sequential) |
| §11 requesting-code-review | PARITY + INTEGRATED | §10 (review as evidence) | Subsumes |
| §12 finishing-a-development-branch | ARCH-DIFF + INTEGRATED | §5 (project uses master-direct instead of branch workflow) | Independent |
| §14 writing-skills | PARTIAL + INTEGRATE-PARTIAL | §10 (durable skills vs disposable skills) | Extends (adds pressure-scenario testing) |
| §15 MMA cluster | PARITY + INTEGRATED | §3 (durable-work + disposable workers + explicit interfaces are MMA's design principle) | Subsumes |
| §16 Dual-Convention | GAP + INTEGRATE-PARTIAL | §10 (single-convention principle; nagent uses one file per concept) | Contradicts (project has 2 conventions) |
## The 3 New Candidates the Superpowers Review Adds
[1 paragraph each: written in Phase 7]
1. **Pressure-scenario testing for skills** (HIGH priority, per decisions.md #2). The superpowers `writing-skills` skill establishes that skills should be TDD-tested via pressure scenarios (red-green-refactor). The project's MMA skills (5 in `.agents/skills/`) and the superpowers-style skills (`brainstorming`, `writing-plans`, etc. as applied to the project) have not been tested via pressure scenarios. The deferred rebuild should add pressure-scenario tests for each MMA skill + the project's spec-writing workflow.
2. **Dual-convention resolution** (HIGH priority, per decisions.md #3). The project has 41 specs/plans in `docs/superpowers/` (per the superpowers-plugin defaults) AND 50+ tracks in `conductor/tracks/<id>/spec.md` (per user choice Q4 = A). The two conventions coexist; the deferred rebuild should resolve this by adopting the conductor convention everywhere and migrating the superpowers-plugin files. nagent_review doesn't address this (it focuses on the data/principle level, not the file location).
3. **The "agent vs agent-tool" boundary explicit rule** (MEDIUM priority, partially in §16). The superpowers skill `using-superpowers` distinguishes between agents that are CLIs (vs APIs), and the project's `conductor/workflow.md` §0 makes the meta-tooling vs application distinction. nagent_review addresses this implicitly via §2 (the Application vs Meta-Tooling section); the superpowers review surfaces it as an explicit "explicit boundary rule" candidate for the deferred rebuild.
## The 2 Nagent Candidates the Superpowers Review Contradicts
[1 paragraph each: written in Phase 7]
1. **The "single source of truth" principle.** nagent_review §10 cites nagent's principle that there should be a single canonical place for each artifact. The superpowers review §16 finds the project has TWO canonical places for specs/plans (`docs/superpowers/` + `conductor/tracks/<id>/`). The contradiction is a project-level violation of nagent's principle; the deferred rebuild should resolve this.
2. **The "YAGNI ruthlessly" principle.** nagent_review §3 cites nagent's "YAGNI ruthlessly" as a load-bearing principle. The superpowers review's HIGH-priority recommended changes (3 items) might be over-engineering for the project's actual needs. The deferred rebuild should re-evaluate the 3 HIGH items against the YAGNI lens.
## Pointer to Fable Review
[1 paragraph: written in Phase 7]
The fable_review (`conductor/tracks/fable_review_20260617/`) analyzes Anthropic's "Claude Fable" system prompt against Manual Slop. Its 13 sections + decisions.md identify ~17 conservative changes for the deferred rebuild. The superpowers review's 25 decisions overlap with fable_review on:
- **§14 of fable** ("Anti-User Watchdog Patterns"): the project rejects Fable's persona-first approach; superpowers §1 (using-superpowers) confirms this via the project being PARITY (no persona engineering).
- **§3 of fable** ("Fable's Refusal Architecture"): the project has no equivalent; superpowers §15 (MMA cluster) shows the project's Tier 4 QA is for diagnostics, not refusal.
- **§13 of fable** ("Genuinely Useful Patterns"): the project's evidence-based discipline; superpowers §4-§6 (TDD, verification, debugging) confirm this.
For the deferred rebuild: read fable_review alongside this track's `decisions.md` to get the full picture. The three reviews (nagent + fable + superpowers) are the user's deferred-rebuild input set.