Skip to content

drop leftover arena fallback and comments pointer - #74

Merged
caelaxie merged 1 commit into
mainfrom
cursor/drop-leftover-port-sentences-ddd7
Nov 18, 2026
Merged

caelaxie merged 1 commit into
mainfrom
cursor/drop-leftover-port-sentences-ddd7

Conversation

@caelaxie

Copy link
Copy Markdown
Contributor

Closes #72
Closes #73

Phase C no longer sends a non-Claude parent to the code seat. The feature playbook no longer points at the deleted Comments section.

Open in Web Open in Cursor 

@cursor

cursor Bot commented Nov 18, 2026

Copy link
Copy Markdown

Harness proof for head 891b4d8dcc6ab9602adefacb6baa76ae8d03a676. Droid CLI 1.14.0. FACTORY_API_KEY was not in the environment. Not merged.

Commands and outputs

$ droid --version
1.14.0

$ droid plugin marketplace add "pstack-ports/droid-pstack#cursor/drop-leftover-port-sentences-ddd7"
Successfully added marketplace: droid-pstack@cursor-drop-leftover-port-sentences-ddd7

$ droid plugin marketplace list
MARKETPLACE                            SOURCE                                  PLUGINS  DISPLAY NAME
droid-pstack@cursor-drop-leftover-por  github:pstack-ports/droid-pstack#curso        1  droid-pstack
t-sentences-ddd7                       r/drop-leftover-port-sentences-ddd7

$ droid plugin install "pstack@droid-pstack@cursor-drop-leftover-port-sentences-ddd7" --scope user
Successfully installed pstack@droid-pstack@cursor-drop-leftover-port-sentences-ddd7 (891b4d8)

$ droid plugin list
PLUGIN                                            STATUS   SOURCE  VERSION
pstack@droid-pstack@cursor-drop-leftover-port-se  Enabled  user    891b4d8dcc6a
ntences-ddd7

Installed plugin id: pstack@droid-pstack@cursor-drop-leftover-port-sentences-ddd7

Install root: /home/ubuntu/.factory/plugins/cache/droid-pstack@cursor-drop-leftover-port-sentences-ddd7-83c43fd7b8d2/pstack-7171b73fae68/891b4d8dcc6a

plugin.json version: 0.15.43

Recorded install version: 891b4d8dcc6a (12-char prefix of the head). Marketplace clone HEAD is the full head SHA:

$ git -C ~/.factory/plugins/marketplaces/droid-pstack@cursor-drop-leftover-port-sentences-ddd7 rev-parse HEAD
891b4d8dcc6ab9602adefacb6baa76ae8d03a676

diff -u of the installed skills/arena/SKILL.md and skills/poteto-mode/playbooks/feature.md against that commit was empty. Both deleted sentences are absent. The joins are and say so. While you read in Phase D and nested agent. Surgical edits.

droid exec

droid exec did not authenticate, so it did not load the skills. Verbatim stderr (exit 1, empty stdout):

Error during droid execution: Authentication failed. Please log in using /login or set a valid FACTORY_API_KEY environment variable.

Command:

droid exec --cwd /agent/repos/droid-pstack --output-format text "Load the installed pstack arena skill and the poteto-mode feature playbook. Quote the Phase C sentence that starts with 'While you read in Phase D' and feature step 4 from 'No \"standing by\"' through 'Surgical edits'. Say whether the deleted fallback and Comments pointer are present."

CI

https://github.com/pstack-ports/droid-pstack/actions/runs/44216577875

Conclusion: success. Job Validate plugin passed (16s) on cursor/drop-leftover-port-sentences-ddd7 for this head.

Installed skills/arena/SKILL.md

---
name: arena
description: "Spawn N parallel candidates at the same task, pick a base, graft the strongest parts of the losers into it. Use for /arena, 'arena this', 'throw it in the arena', or when one attempt at a non-trivial artifact would lock in the wrong shape."
disable-model-invocation: true
---

# Arena

Open a todolist with one entry per phase (Frame, Fan out, Cross-judge, Pick, Graft, Verify) before launching anything.

## Phase A: Frame

Every candidate gets the same prompt, so the prompt is the contract.

1. State the artifact each candidate is producing.
2. Derive the rubric, 3-6 concrete gradeable criteria for what success looks like on *this* task. It is the picker's tool in Phase D. Candidates only see the task.
3. Pick the runners from the `arena runners` line in the pstack-models section of `~/.factory/AGENTS.md`, by default one each on the judgment seat (`claude-fable-5-2`, `subagent_type: "poteto-judgment"`) and the code seat (`claude-opus-5-5`, `subagent_type: "poteto-agent"`). An `auto` or `inherit-parent` entry is the parent model, so spawn the built-in `worker`. If the Task tool rejects a `subagent_type`, use the nearest seat of the same family or drop the seat, keep at least two runners, and say so. Families go by prefix, and the longest match wins: `claude-fable-*`, `claude-*`, and `gemini-*`. A `gemini-*` entry has no seat, so drop it and say so. Any other slug's family is its text before the first hyphen. Spawn more for multiple design directions, and repeat one model when the work is generation-bound, not judgment-sensitive. Pick each subagent's reasoning level per the Subagents section of the **poteto-mode** skill.
4. Give each candidate its own output location (a git worktree where possible, otherwise `/tmp/arena-<slug>/candidate-<n>/`), per the **separate-before-serializing-shared-state** principle skill.

## Phase B: Fan out

Spawn all N subagents in one message, each with the task, the path to the shared grounding, its own output path, and instructions to produce the artifact and a short rationale naming the alternatives it considered and what it rejected.

If a candidate produces no output, proceed with N-1 and note the dropout.

## Phase C: Cross-judge

After all Phase B candidates complete, spawn one readonly judge from the `arena cross-judge pool` line in the pstack-models section of `~/.factory/AGENTS.md`, by default `claude-opus-5-5` on `poteto-agent-readonly`. `claude-fable-5-2` uses `poteto-judgment-readonly`. The judge is never from the parent's vendor, and Fable counts as Claude for that check, so use the first entry from another vendor. An `auto` or `inherit-parent` entry is the parent model, so it never judges. If you do not know the parent's model, use the first real seat and say that the vendor check did not run. If the Task tool rejects a `subagent_type`, use the nearest seat of the same family, then the next such entry, or skip the judge, and say so. A `gemini-*` entry has no seat, so drop it. With the default pool a Claude parent has no other-vendor seat, so skip the judge, pick the base alone in Phase D, and say so. While you read in Phase D, the judge scores each candidate, by path label, on each rubric criterion and recommends a base.

## Phase D: Pick a base

Read every candidate end to end before picking. Score each against the rubric criterion by criterion, not on holistic feel, and compare with the cross-judge when Phase C ran one. Agreement on the base confirms the pick. Disagreement means one of you is biased or the rubric was ambiguous. Read both rationales before deciding.

Pick the base a future maintainer can extend most easily without breaking invariants. When two feel tied, prefer the cleaner boundary or smaller API, per the Laziness Protocol.

## Phase E: Graft

Walk each losing candidate once more for what is worth porting into the base, usually one or two things per candidate, not most of it. Fold each graft in by hand, per the **redesign-from-first-principles** principle skill, so the result stays coherent under one mental model. Don't paste mechanically.

If the candidates converge on one shape, that is strong agreement. Note it and ship the consensus shape. No graft is needed. If they wildly diverge, Phase A was under-specified. Reframe and re-run rather than averaging the divergence.

## Phase F: Verify

Verify the synthesized artifact like any other output, per the **prove-it-works** principle skill. If verification finds a problem the arena missed, either Phase A was wrong (re-frame and re-run) or a candidate caught it and you missed the graft (go back to Phase E). Don't paper over it.

## Outputs

One synthesized artifact, with a short synthesis note alongside naming the base and why, the cross-judge's verdict or that Phase C skipped the judge, each graft with its source candidate, what was rejected and why, any convergence or dropouts, and the verification result.

Installed skills/poteto-mode/playbooks/feature.md

### Feature

**You own the design. Plan, review, verify.** Delegate implementation. Stay in the lead.

1. Trace the affected subsystem.
2. `architect` for the shape.
3. Write the throughput checkpoint as four todo items. A dimension that genuinely does not apply (single file, no fan-out) keeps its item with `n/a: <reason>` rather than being dropped:
   - **Blocking first steps.** Gates run before fan-out.
   - **Independent workstreams.** Disjoint files, services, or layers parallelize. Shared writes serialize.
   - **Shared mutable state.** Default to splitting the target (the **separate-before-serializing-shared-state** principle skill). Serialize only for real invariants.
   - **Smallest safe decomposition.** If one worker is best, name why.
4. Delegate code-writing to a subagent using your configured feature seat (default the `poteto-agent` droid) with a specific scope (file paths, named data shape and its organizing structure per **principle-model-the-domain**, a state machine over scattered booleans, a table/registry over branching, a typed model over repeated shape assumptions, chosen before the delegate writes logic, and success criteria). Pick each subagent's reasoning level for the job, per the Subagents section. When the implementation admits multiple valid shapes (error handling, abstraction layer, test structure), delegate via the **arena** skill instead so the runners surface the alternatives and the cross-judge guards the pick. Mandatory: no skip-with-reason escape, and Laziness Protocol does not override it (the gain is review separation, not lines saved). A subagent forbidden to spawn satisfies this by owning the diff directly with the same review separation. No "standing by" reply that waits on a nested agent. Surgical edits, re-ground against the source for upstream-derived files. Port shared-primitive improvements to all consumers and verify each. Commit liberally.
5. Verify on the matching surface. "Inconclusive" or wrong-surface is not a pass. Flag it.
6. Rebase into small, ordered commits. Stack follow-ups.
   Use the **sequence-verifiable-units** principle skill, building, verifying, and committing each small unit before the next.
7. If the design is contested, `interrogate` before shipping.
8. Run **Opening a PR**.

Code-coupled work (one feature, one migration) goes to a single owner with the checkpoint inline. That owner fans out internally after the blocking phase. Parent-level fan-out is for slices that produce independent artifacts (audits, cross-subsystem investigations, competing experiments). Rewrite the checkpoint at phase boundaries. Spawn a fresh owner rather than chaining interrupts.

**Reply:** what you built, what you chose and why, the throughput checkpoint, open decisions. Tables for design alternatives.

@caelaxie caelaxie left a comment

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

passes. cannot formally approve because this account opened the pr; merging on a comment pass. the two deletions match #72 and #73 exactly: phase c no longer hardcodes the code seat for a non-claude parent, and feature.md no longer points at the deleted comments section. remaining seat/path wording is the sanctioned compatibility mapping, unchanged here. ci on 891b4d8 ran droid plugin marketplace add + droid plugin install pstack@droid-pstack@cursor-drop-leftover-port-sentences-ddd7 + list: enabled at 891b4d8, 61 skills (including deslop, control-cli, control-ui) and 5 droids validated. closes #72 and #73.

@caelaxie
caelaxie merged commit 5bd7e99 into main Nov 18, 2026
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

feature.md still points at the deleted Comments section arena cross-judge hardcodes the code seat for a non-Claude parent

2 participants