Skip to content
Merged
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
22 changes: 19 additions & 3 deletions .github/workflows/ci.yml
Original file line number Diff line number Diff line change
Expand Up @@ -28,6 +28,16 @@ jobs:
- name: Install Dependencies
run: npm install

# Installed before the test suite below because some root tests (for example
# tests/control-panel-work-item-writer.test.mjs) import apps/control-panel/
# server/*.mjs directly, which needs js-yaml -- a dependency that lives only in
# apps/control-panel/node_modules. This only touches design-system/ and
# apps/control-panel/, neither of which the base-branch checkouts below ever
# overwrite, so it doesn't affect the trust boundary those checkouts establish
# for the gate scripts.
- name: Install control-panel frontend dependencies
run: npm install --prefix design-system && npm install --prefix apps/control-panel

# Snapshot freshness, the test suite, and AgentProof all run BEFORE the
# base-branch checkouts below, which overwrite the gate scripts and the
# attribution detector kernel with base versions. All three need to see
Expand Down Expand Up @@ -167,12 +177,12 @@ jobs:
- name: Run work item validation gate
run: node scripts/check-work-items.mjs

- name: Run work item staleness gate
run: node scripts/check-work-item-staleness.mjs

- name: Run checker engagement gate
run: node scripts/check-checker-engagement.mjs

- name: Install control-panel frontend dependencies
run: npm install --prefix design-system && npm install --prefix apps/control-panel

- name: Run design-system build
run: npm run build --prefix design-system

Expand Down Expand Up @@ -200,6 +210,12 @@ jobs:
- name: Install Dependencies
run: npm install

# See the matching comment in the ratchet job above: some root tests import
# apps/control-panel/server/*.mjs, which needs js-yaml from that workspace's
# own node_modules.
- name: Install control-panel frontend dependencies
run: npm install --prefix design-system && npm install --prefix apps/control-panel

- name: Run drift guard gate
run: node scripts/check-drift.mjs

Expand Down
42 changes: 42 additions & 0 deletions .github/workflows/modonome-review.yml
Original file line number Diff line number Diff line change
@@ -0,0 +1,42 @@
# Modonome independent checker on every pull request (ADR-038 spike).
#
# The maker/checker loop's checker is not private to the internal loop: it can review
# any change, whoever produced it (a human, a coding-agent session, or the internal
# maker). This workflow runs that review-only checker over the PR diff and writes the
# result to the job summary, so the change gets an independent adversarial review before
# merge regardless of origin.
#
# Deliberately inert for the spike: it runs a DRY-RUN (no model call), needs no secrets,
# and holds contents:read only. A live review that calls a local or free checker and
# posts a PR review is the ADR-038 follow-up; it needs a configured checker endpoint and
# a write-scoped token, so it is not enabled here.
name: Modonome Review

on:
pull_request:
branches: [main]

permissions:
contents: read

jobs:
review:
runs-on: ubuntu-latest
steps:
- name: Check out the PR head with history
uses: actions/checkout@v4
with:
fetch-depth: 0

- name: Set up Node
uses: actions/setup-node@v4
with:
node-version: "20"

- name: Independent checker review (dry-run)
run: |
git diff "origin/${{ github.base_ref }}...HEAD" > /tmp/pr.diff
node scripts/agent/review-diff.mjs \
--diff /tmp/pr.diff \
--author "PR #${{ github.event.pull_request.number }} by @${{ github.event.pull_request.user.login }}" \
>> "$GITHUB_STEP_SUMMARY"
27 changes: 19 additions & 8 deletions .modonome/STATUS.md
Original file line number Diff line number Diff line change
Expand Up @@ -6,14 +6,25 @@

## Queue

The work queue is empty. Modonome's own implementation is stable. See `ROADMAP.md` for
the committed roadmap (Milestones 1-6) and `docs/research/` for exploratory directions.

**Note on historical work items:** The initial `.modonome/work-items/` queue contained
18 seeded example items (WI-001 through WI-018) that were part of Modonome's development.
These items have all been implemented and merged (visible in git history) but were left
as example artifacts. They are no longer active. See `examples/demo-app/.modonome/` for
an active demonstration of Modonome's governance workflow on a real application.
This section is a snapshot, not the source of record: `.modonome/work-items/` on disk
always wins. Run `node scripts/check-work-items.mjs` for a live pass/fail, or
`node scripts/check-work-item-staleness.mjs` to catch an item whose state has drifted
from what actually landed (see the note below on why that check exists).

As of this snapshot: 36 of 39 items are `done`. Three are genuinely open:
`WI-020` (validate-work-item error-message improvements), `WI-032` (configurable
trigger and orchestration layer), and `WI-033` (GitHub Models workflow wiring). See
`docs/autonomy-plan.md` for the dependency-ordered plan those last two belong to.
See `ROADMAP.md` for the committed roadmap (Milestones 1-6) and `docs/research/` for
exploratory directions.

**Note on state drift:** This file previously claimed the queue was empty and only
acknowledged WI-001 through WI-018 as historical, while WI-019 through WI-040 had
been added and 15 of them (WI-021, WI-022, WI-026 through WI-031, WI-034 through
WI-040) were already merged with passing gates but left marked `queued`/`claimed`.
Nothing had reconciled `state:` against merged reality, since `transition-work-item.mjs`
is a manual step and no CI gate checked for this. `check-work-item-staleness.mjs` now
runs in `npm run verify` specifically to catch a repeat of this.

## Decisions

Expand Down
10 changes: 10 additions & 0 deletions .modonome/config.yaml
Original file line number Diff line number Diff line change
Expand Up @@ -67,6 +67,16 @@ roles:
self-govern:
runner: container
model: claude-haiku-4-5-20251001
# Researcher agent: a capability profile example (ADR-039). It is NOT on the active
# role_sequence (maker -> checker), so it does not run in the loop yet; it is defined
# here so its skills, tools, and prioritized model fallback are configurable ahead of
# loop wiring. models is a priority list: the paid frontier choice first, falling back
# to the free/local one when no daily budget is set.
researcher:
runner: container
models: [claude-opus-4-8, local-default]
skills: [cite-sources, summarize-findings]
tools: [web-search, read-only-fs]
runners:
local:
labels: [self-hosted, mac-mini]
Expand Down
3 changes: 2 additions & 1 deletion .modonome/policy-attestation.json
Original file line number Diff line number Diff line change
Expand Up @@ -98,6 +98,7 @@
"check-repo-hygiene",
"check-self-application",
"check-style",
"check-work-item-staleness",
"check-work-items",
"policy-attestation-freshness",
"snapshot-freshness",
Expand All @@ -108,5 +109,5 @@
"autonomy_enabled": false,
"dry_run": true
},
"content_digest": "sha256:0aab6fd72545a09182ca221157529a674e861265ddc4a2b8fcf34b17bd48176f"
"content_digest": "sha256:e5b6d8ac37e48a32a7e56b6a1fadb785bcfac9d17e6599fae74a660ab33b0a8b"
}
2 changes: 1 addition & 1 deletion .modonome/snapshot/badge.json
Original file line number Diff line number Diff line change
@@ -1 +1 @@
{"color":"blue","label":"snapshot","message":"845 files, 106536 tok","schemaVersion":1}
{"color":"blue","label":"snapshot","message":"868 files, 113396 tok","schemaVersion":1}
2 changes: 1 addition & 1 deletion .modonome/snapshot/map.json

Large diffs are not rendered by default.

Loading
Loading