Skip to content
Closed
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 0 additions & 6 deletions .agents/skills/effort-graph/release.json

This file was deleted.

File renamed without changes.
6 changes: 6 additions & 0 deletions .agents/skills/proof/release.json
Original file line number Diff line number Diff line change
@@ -0,0 +1,6 @@
{
"format": 1,
"flatbreadVersion": "1.0.0",
"effortGraphVersion": "1.0.0",
"gitTag": "v1.0.0"
}
File renamed without changes.
2 changes: 1 addition & 1 deletion .cursor/agents/flatbread-architecture-planner.md
Original file line number Diff line number Diff line change
Expand Up @@ -31,7 +31,7 @@ Keep recommendations aligned with `flatbread-flow-pmf-audit.md` and `flatbread-f

## Output Schema For DAG Handoff

When invoked inside a DAG task (the `proof` skill / `@flatbread/proof` package), keep the response under ~1800 chars and lead with these exact `##` headings so downstream tasks can find them after the 2000-char upstream stitch cap:
When invoked inside a Proof DAG task (external CLI from https://github.com/FlatbreadLabs/proof, package `@flatbread/proof`), keep the response under ~1800 chars and lead with these exact `##` headings so downstream tasks can find them after the 2000-char upstream stitch cap:

```
## Current contract
Expand Down
16 changes: 10 additions & 6 deletions .cursor/agents/flatbread-code-review-orchestrator.md
Original file line number Diff line number Diff line change
@@ -1,13 +1,13 @@
---
name: flatbread-code-review-orchestrator
description: Orchestrates an adversarial code-review DAG over a Flatbread change diff using `@flatbread/proof` — picks ≤5 reviewer perspectives dynamically from the diff, runs them in a single rank, and merges their findings through a judge that emits chunk-bound structured feedback.
description: Orchestrates an adversarial code-review DAG over a Flatbread change diff using the external Proof CLI (`@flatbread/proof`) — picks ≤5 reviewer perspectives dynamically from the diff, runs them in a single rank, and merges their findings through a judge that emits chunk-bound structured feedback.
readonly: true
tools: ReadFile, Glob, rg, Shell
---

# Flatbread Code-Review Orchestrator

You orchestrate an adversarial code review over a Flatbread change diff. Reviewers run as parallel local subagents under `@flatbread/proof`, then a judge merges their findings into structured feedback bound to specific diff chunks. You do not edit code. You produce a DAG, run it, and report.
You orchestrate an adversarial code review over a Flatbread change diff. Reviewers run as parallel local subagents under the external Proof CLI (`@flatbread/proof` from https://github.com/FlatbreadLabs/proof), then a judge merges their findings into structured feedback bound to specific diff chunks. You do not edit code. You produce a DAG, run it, and report.

Your loyalty is to **correctness** and **wide, robust test coverage**. Pedantry is the failure mode you must avoid (see `## Anti-Pedantry Guardrails`).

Expand Down Expand Up @@ -91,18 +91,23 @@ Pass the diff to each subtask by **file path reference**, not by inlining. The d

### 4. Run the DAG

Proof is an external package (`@flatbread/proof` from https://github.com/FlatbreadLabs/proof). It is not a workspace package in this monorepo. Install it as a root dependency (`pnpm add -Dw @flatbread/proof`) or otherwise ensure `pnpm exec proof` resolves before continuing. Also set `CURSOR_API_KEY` (or load it from `.env`).

```bash
[ -n "$CURSOR_API_KEY" ] || { [ -f .env ] && set -a && source .env && set +a; }

CANVAS_PATH="$HOME/.cursor/projects/$(pwd | sed 's|^/||; s|/|-|g')/canvases/dag-review-$(git rev-parse --short HEAD).canvas.tsx"
pnpm exec proof --help >/dev/null || {
echo "Install @flatbread/proof so pnpm exec proof resolves (pnpm add -Dw @flatbread/proof)."
exit 1
}

[ -f "$(git rev-parse --show-toplevel)/packages/proof/dist/run_dag.js" ] || pnpm -F @flatbread/proof build
CANVAS_PATH="$HOME/.cursor/projects/$(pwd | sed 's|^/||; s|/|-|g')/canvases/dag-review-$(git rev-parse --short HEAD).canvas.tsx"

pnpm exec proof --init-only --dag /tmp/review-dag.json --canvas-path "$CANVAS_PATH"
open "$CANVAS_PATH" >/dev/null 2>&1 || true
```

Then surface the canvas link in chat using the exact text `Open Canvas` (per the `proof` skill), then run for real:
Then surface the canvas link in chat using the exact text `Open Canvas`, then run for real:

```bash
pnpm exec proof --dag /tmp/review-dag.json --canvas-path "$CANVAS_PATH"
Expand All @@ -128,7 +133,6 @@ Each perspective is read-only, scoped, and produces the same handoff schema (`##
| `docs-and-positioning` | LOW | `*.md`, `docs/**`, `flatbread-flow-*.md` | Repositioning drift (Flatbread is _not_ a general flat-file database; see `flatbread-major-migration` skill). Stale commands. Broken cross-links. |
| `release-discipline` | MED | `package.json` version bumps, `scripts/publish.ts`, `CHANGELOG.md`, `pnpm-workspace.yaml` | Coordinated monorepo bumps for breaking changes; prerelease train discipline (`1.0.0-alpha.N` / `-beta.N`); accidental `latest` dist-tag; missing migration notes. |
| `perf-and-caching` | MED | `**/cache*.ts`, `**/hash*.ts`, `**/dependencyCheck*.ts`, hot-path resolver edits | Cache key collisions, unbounded memoization, accidental O(n²) on file count, FS calls in tight loops. |
| `proof-runtime-internals` | HIGH | `packages/proof/src/**` | DAG topo-sort correctness, stream throttling/idle-timeout edges, supervisor exit-code (75) contract, canvas-write atomicity, `--restart-on-runner-change` boundaries. |

You may add a one-off bespoke perspective if the diff's center of gravity is outside this catalog (e.g. a new `packages/<thing>/`), but keep the cap at 5 total.

Expand Down
46 changes: 0 additions & 46 deletions .cursor/agents/flatbread-proof-runtime-skeptic.md

This file was deleted.

40 changes: 40 additions & 0 deletions .cursor/dags/README.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,40 @@
# Flatbread Proof DAGs

Flatbread-specific DAG JSON for the external Proof CLI
([`@flatbread/proof`](https://github.com/FlatbreadLabs/proof)). These files
belong here, not in the Proof repo: their prompts call monorepo commands such
as `pnpm --filter @flatbread/…` and name Flatbread packages, ports, and agents.

## Layout

- `flatbread/` — workspace orchestration templates
- `dag-schema-migration.json` — schema-breaking migration (21 tasks)
- `dag-codegen-change.json` — codegen-focused change
- `dag-docs-sync.json` — docs/positioning sync
- `dag-flatbread-flow-pmf-audit.json` — PMF audit flow

## Prerequisites

1. Install Proof so `pnpm exec proof` resolves from this repo root, for example:

```bash
pnpm add -Dw @flatbread/proof
```

Or install from the Proof repository if the package is not on the registry
you use yet.

2. Set `CURSOR_API_KEY` (or load it from `.env`).

## Run a DAG

From the Flatbread repo root:

```bash
pnpm exec proof --init-only --dag .cursor/dags/flatbread/dag-schema-migration.json
pnpm exec proof --dag .cursor/dags/flatbread/dag-schema-migration.json
```

Replace the DAG path with any file under `flatbread/`. Edit task prompts and
`depends_on` before a real run; `--init-only` confirms the rank shape without
spending model calls.
Original file line number Diff line number Diff line change
Expand Up @@ -46,7 +46,7 @@
"id": "recommend-roadmap",
"depends_on": ["synthesize-pmf-gaps"],
"complexity": "HIGH",
"subtask_prompt": "You are acting as `flatbread-architecture-planner` producing a roadmap recommendation. Follow its output schema. Output must lead with these `##` headings verbatim: `## Current contract`, `## Proposed contract`, `## Migration impact`, `## Validation plan`, `## Human checkpoints`.\n\nBased on the synthesized PMF gaps, recommend a concise product direction and roadmap. Output a sharper positioning statement, 3-5 product primitives to add or clarify (each with file/package anchor), near-term experiments, and what not to build yet. `## Migration impact` should map each recommendation to the affected packages so a follow-up `flatbread-major-migration` DAG (template at `.cursor/skills/proof/examples/flatbread/dag-schema-migration.json`) can be authored from this output without re-deriving scope."
"subtask_prompt": "You are acting as `flatbread-architecture-planner` producing a roadmap recommendation. Follow its output schema. Output must lead with these `##` headings verbatim: `## Current contract`, `## Proposed contract`, `## Migration impact`, `## Validation plan`, `## Human checkpoints`.\n\nBased on the synthesized PMF gaps, recommend a concise product direction and roadmap. Output a sharper positioning statement, 3-5 product primitives to add or clarify (each with file/package anchor), near-term experiments, and what not to build yet. `## Migration impact` should map each recommendation to the affected packages so a follow-up `flatbread-major-migration` DAG (template at `.cursor/dags/flatbread/dag-schema-migration.json`) can be authored from this output without re-deriving scope."
}
]
}
32 changes: 0 additions & 32 deletions .cursor/skills/dag-task-runner/SKILL.md

This file was deleted.

4 changes: 2 additions & 2 deletions .cursor/skills/flatbread-major-migration/SKILL.md
Original file line number Diff line number Diff line change
Expand Up @@ -64,7 +64,7 @@ Treat breaking work as a coordinated monorepo release across `flatbread`, `@flat

## Default DAG Shape

When this skill is run under `proof` (the `@flatbread/proof` package; legacy alias `dag-task-runner`), use the topology in `flatbread-flow-agentic-workflows.md` ("DAG Topology" section). The canonical schema-migration shape — express the DAG via `depends_on` only; the runner computes ranks via Kahn topo-sort. The shape below is what `pnpm exec proof --init-only` produces for the starter template (21 tasks across 7 ranks):
When this skill is run under the external Proof CLI (`@flatbread/proof` from https://github.com/FlatbreadLabs/proof), use the topology in `flatbread-flow-agentic-workflows.md` ("DAG Topology" section). The canonical schema-migration shape — express the DAG via `depends_on` only; the runner computes ranks via Kahn topo-sort. The shape below is what `pnpm exec proof --init-only` produces for the starter template (21 tasks across 7 ranks):

```
rank 1 diag-schema, diag-resolvers, diag-types, diag-codegen, diag-cli,
Expand All @@ -79,7 +79,7 @@ rank 6 verify-cli (reviewer; b
rank 7 browser-verify (browser-verifier; binds port 5057, depends_on impl-examples + verify-cli)
```

Two ranks bind port `5057` (`verify-cli`, `browser-verify`); they must remain on distinct ranks. A starter JSON lives at `.cursor/skills/proof/examples/flatbread/dag-schema-migration.json`. Edit task contents and `depends_on`, then re-run `--init-only` to confirm the rank shape didn't regress.
Two ranks bind port `5057` (`verify-cli`, `browser-verify`); they must remain on distinct ranks. A starter JSON lives at `.cursor/dags/flatbread/dag-schema-migration.json`. Edit task contents and `depends_on`, then re-run `--init-only` to confirm the rank shape didn't regress. See `.cursor/dags/README.md` for how to install and run Proof against these DAGs.

### Safe parallel cuts

Expand Down
Loading
Loading