Two follow-ups to the server/src env-var scan widening:
1. verify-edit.mjs had its own copy of shouldFetchEnvSource with the same
narrow server allowlist — apply the identical server/src/**/*.ts widening
so the verify phase stops emitting unverified env-var false positives.
2. Both scripts now also scan all non-test packages/adapters/<name>/src/**/*.ts.
Adapter packages read env vars directly in their server/index modules —
e.g. SANDBOX_INSTALL_COMMAND (packages/adapters/{pi,claude,gemini,cursor}-local/src/...)
and per-runtime API-key fallbacks — which the old allowlist never fetched.
Verified against parent v2026.618.0:
- check-drift: 0 drift (65 env vars checked).
- verify-edit docs/reference/adapters/sandbox-providers.md: 6/6 claims verified
(SANDBOX_INSTALL_COMMAND, previously unverified, now resolves).
- sync:test 33/33.
Selection grows to ~452 files, still well under the 700 cap.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
The env-var drift checker only fetched a narrow allowlist of server files
(config.ts, a few auth/service files), so env vars read elsewhere in
server/src showed up as env-var-missing false positives every run.
Verified against parent v2026.618.0 that the 5 recurring false positives are
real, read env vars the old scan never looked at:
- TRUST_PROXY server/src/app.ts:167 (parseTrustProxyEnv(process.env.TRUST_PROXY))
- OTEL_EXPORTER_OTLP_ENDPOINT server/src/instrumentation.ts:29
- OTEL_EXPORTER_OTLP_PROTOCOL server/src/instrumentation.ts:69
- OTEL_SERVICE_NAME server/src/instrumentation.ts:144
- OTEL_SERVICE_VERSION server/src/instrumentation.ts:145
Fix: replace the three narrow server/src rules in shouldFetchEnvSource with a
single rule covering every non-test server/src/**/*.ts file (tests already
excluded above). Selection grows 130 → 326 files, well under the 700 cap.
Result: check-drift against both v2026.618.0 and master now reports 0 drift
(was 5 env-var-missing), with identical coverage (65 env vars, 235 routes,
57 CLI commands, 7 paths). sync:test 33/33 pass.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
The "Bundled with the Paperclip server (skills/)" list claimed three skills
that are NOT server-bundled — they live under .agents/skills/ (repo-development
skills) at v2026.618.0 and also at v2026.609.0, so this is pre-existing drift.
Verified against parent v2026.618.0:
- server-bundled skills/ contains exactly 6 dirs (all force-imported + required
by ensureBundledSkills): paperclip, paperclip-board, paperclip-create-agent,
paperclip-dev, paperclip-converting-plans-to-tasks, para-memory-files.
- paperclip-create-plugin, diagnose-why-work-stopped, terminal-bench-loop are
under .agents/skills/, not skills/.
Fixes:
- Remove the 3 non-bundled skills from the bundled list (now matches skills/ 1:1).
- Drop the brittle "five paperclip-*" count (it also wrongly excluded the bundled
para-memory-files); phrase count-free.
- Repair the paperclip-create-plugin link: parent has no `main` branch and the
skill moved, so blob/main/skills/... (404) -> blob/master/.agents/skills/...
verify-edit: 7/7 claims verified. sync:check clean (no broken internal links).
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Cumulative window v2026.609.0 → 1b89a8e. The only commit added since the
last run (412a04c → 1b89a8e) is a Dockerfile fix removing VOLUME
["/paperclip"] for Railway compatibility — untracked by any watcher and
with no doc impact (the fly.toml volume still mounts at /paperclip).
Drift: 5 records (TRUST_PROXY, OTEL_*) all false positives — real server
env vars the drift checker can't see because it doesn't scan server/src
process.env reads. Confirmed present upstream; documented in the prior run.
No edits applied. Refreshed PENDING.md / SCREENSHOTS_PENDING.md headers
and bumped last_seen_parent_sha + last_run_outcome=no-changes.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
- The per-project Workspaces tab is hidden until the project has workspace
activity (showWorkspacesTab requires summaries > 0); say so instead of
'shows up on each project', and drop the never-shown per-project empty
string. The reliable empty state is the company-wide 'No workspace
activity yet.'
- Correct the surfacing rule: a primary workspace surfaces because it's the
primary (not 'has a working directory and runtime config'); non-primary
project workspaces need runtime services/config.
- Replace the stale 'Each entry shows' list (old table columns: mode /
provider type / opened) with the actual ProjectWorkspaceSummaryCard fields
(kind, name, branch, path, services, status, linked tasks, updated).
- Note plugin tabs appear alongside the base tabs (e.g. workspace-diff's
Changes tab), not strictly after them.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
A user reported the Workspaces screen looked empty with "no options," and
that all but one of the workspace screenshots were blank/identical.
Both were real problems:
Docs (docs/guides/projects-workflow/workspaces.md)
- Add a "Before you start" section: workspaces are gated behind the
experimental "Enable Isolated Workspaces" flag, the screen is monitor-only
(no manual create), and it stays empty until an agent runs an isolated
task — with the actual empty-state strings ("No workspace activity yet." /
"No non-default workspace activity yet.").
- Correct the detail screen: it now has Tasks / Services / Configuration /
Runtime logs / Routines tabs (+ plugin tabs), not "three tabs." Rename the
Issues tab to Tasks (its real label) and add a Services tab section. The
former "Workspace commands panel above the tabs" is now the Services tab
plus header quick controls.
- Clarify the company-wide Workspaces screen vs. the per-project Workspaces
tab and the surfacing rules. Fix the misleading "multiple isolated
workspaces" caption.
Screenshots
- Root cause: the seed only created a *project* workspace and pointed the
execution-workspace detail routes at its id, so those pages were captured
mid "Loading workspace…". Execution workspaces have no REST create endpoint
(they're provisioned by a real heartbeat), so seed-execution-workspace.mjs
now inserts a realistic one directly into the instance's embedded Postgres:
an isolated git-worktree workspace with runtime config, a running service,
provision/start/failed-job operation logs, and linked tasks.
- routes.mjs: add the {executionWorkspaceId} token, repoint the detail routes
to it, add a Services tab target, and drop the duplicate
org/execution-workspaces-list target (same URL as workspaces/list).
- Recaptured workspaces/{list,configuration,services,runtime-logs,issues} in
light + dark; all now render real content.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
main shipped v2026.609.0 (PR #25). Advance nightly's cumulative-diff base
to v2026.609.0, switch branch_mode back to nightly, clear the applied
manifest hash, and reset PENDING.md so the next nightly run starts from a
clean v2026.609.0 baseline.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Publish accumulated nightly drafts to main and document the new surfaces
shipped between v2026.529.0 and v2026.609.0 (130 parent commits).
New pages (gaps not covered by nightly drafts):
- docs/reference/cli/token.md — `token agent` / `token board` API-key commands
- docs/reference/cli/connect.md — interactive `connect` setup wizard
- docs/reference/api/teams-catalog.md — teams catalog REST API
Also:
- Stamp paperclip_version: v2026.609.0 on 49 release doc pages
- Register the 3 new pages in site/content.json nav
- Bump .sync-state.json base_release_tag to v2026.609.0
Verification: 49 pages checked against v2026.609.0; 11 verifier flags,
all confirmed false-positives (flagged flags/env-vars exist in parent).
Drift: 0. Reconciliation: 0.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
- docker.md / gemini-local.md: Gemini CLI now bundled in the official
image (GEMINI_SANDBOX=false); backfill opencode_local in bundled list
(paperclip #7693)
- external-adapters.md: external adapters can override built-in types;
document override + fallback lifecycle, correct stale 'cannot be
overwritten' note (paperclip #7394)
- regenerate PENDING.md / SCREENSHOTS_PENDING.md (18 settings/adapter
screenshots stale after instance-settings IA move, paperclip #7680)
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
`company current` exists in the CLI source
(cli/src/commands/client/company.ts) but was missing from the company
reference page. Add it to the Read Commands code block and table with a
note on how it resolves the scoped company (--company-id, context,
PAPERCLIP_COMPANY_ID, or agent credential) and that it returns the same
full record as `company get <companyId>`.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Three settled surfaces aged past the 24h quarantine since the last run:
- Teams Catalog (#7550): new `teams` CLI page + Team Catalog guide for the
installable bundled/optional starter teams.
- Company Artifacts (#7621): new Artifacts guide + `GET /api/companies/{id}/artifacts`
API section under Companies.
- Low-trust review containment / trust presets (#7530): new Administration page
covering the `standard` / `low_trust_review` presets and quarantine-then-promote
review flow.
Nav merged in site/content.json. PENDING.md and SCREENSHOTS_PENDING.md regenerated
(40 screenshots stale in the settled window). Phase 5.5 verification clean; drift clean.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Cumulative window v2026.529.0 → e78c064 (24h-quarantined nightly cutoff).
Three user-visible surfaces newly documented this run:
- skills: app-shipped catalog now supports *referenced* entries fetched from
an external GitHub repo pinned to a commit (CatalogSkill.source). First one
is the research-category last30days skill. Documented in the Skills
reference and the org Skills guide.
- routines: scheduled ticks are suppressed (recorded as skipped, no backfill)
while the routine's project is paused.
- companies: archiving runs a cascade that pauses active agents
(pauseReason company_archived) and cancels queued wake-ups.
All claims verified against parent code (Phase 5.5 clean). Build + link/nav
checks pass; no drift. Teams Catalog, company-artifacts page, and low-trust
review containment are quarantined (<24h) and deferred to the next run.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Closes the gap this run surfaced: packages/skills-catalog/** was unwatched, so
new bundled catalog skills (e.g. the wireframe product skill) weren't auto-detected.
Watches catalog/**/SKILL.md + generated/catalog.json, routes to docs/reference/skills.md
and docs/guides/org/skills.md (PR tier), distinct from the agent-skills watcher which
covers the server-forced bundled skills under skills/**.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Released v2026.529.0 docs to main (PR #19). Advance the cumulative-diff
base so the next nightly run diffs from this release, not v2026.525.0.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Recapture the 82 stale screenshots (light + dark, 1440×900 @2x) against
a release-code instance booted at v2026.529.0 (911a1e8). Registry was
already stamped at that SHA, so captured_sha is unchanged.
Verification pass against the release codebase also found one authored
claim wrong: the execution-workspace environment precedence is
issue → project → agent → instance default (project policy outranks an
agent's own default), not issue → agent → project. Corrected
docs/guides/projects-workflow/workspaces.md to match
server/src/services/execution-workspace-policy.ts.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Add the two release features quarantined from nightly, and stamp
paperclip_version on every touched page.
- Claude Local: document UI-driven live model discovery (#6953) —
/v1/models lookup via ANTHROPIC_API_KEY, 60s cache, built-in fallback,
Bedrock IDs, and the refresh control.
- Workspaces: document reused-workspace environment consistency and
finalize-gated dependent wakes / interaction.accept (#6969).
- Stamp paperclip_version: v2026.529.0 on all 14 touched pages.
- Regenerate SCREENSHOTS_PENDING.md: 90/276 stale for this window.
Inherits the six features already drafted on nightly: resource
memberships (#6677), document annotations (#6733), bundled plugins in
plugin manager (#6734), skills CLI + catalog (#6782), first-admin
claim (#6755).
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
The seeded org agents were created as `process` adapters with an empty
adapterConfig. The screenshot instance runs with no LLM provider and the
scheduler heartbeats every enabled agent, so each agent's heartbeat failed
instantly — the captured UI was dominated by red "error" states ("5 errors"
on the dashboard, "Ada failed after 1 second" on agent detail).
Give every demo agent a shared DEMO_AGENT_CONFIG: a process command that
always exits 0. Heartbeats now complete cleanly, so agents render as healthy
idle/running/paused with tidy transcripts. Recaptured all targets (light +
dark) against paperclip 911a1e8 — dashboard now reads "0 errors".
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
New scripts/sync/verify-screenshots.mjs asserts that every screenshot
referenced in a guide page is (a) a CAPTURE_TARGET in routes.mjs and (b)
indexed in registry.json with a non-empty depends_on. It exits non-zero
otherwise, so a doc image added without a capture target fails loudly instead
of silently going stale. Wired into `npm run sync:check`, which the /sync-docs
skill runs every sync.
Currently: 112 referenced screenshots, all tracked and indexed.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Previously sync-registry only *updated* existing registry entries, so a newly
introduced screenshot target was captured but never indexed — and the
/sync-docs staleness check only flags entries it can see, so such a shot would
silently drift out of date (exactly what happened to issues/inbox-unread).
sync-registry now appends a registry entry (file, theme, viewport, route,
depends_on) for any CAPTURE_TARGET that has none, and sorts entries for stable
diffs. run.mjs runs sync-registry on every capture, so the index now stays
complete automatically. Audit: 0 target files missing from the registry, 0
without depends_on — every documented screenshot is tracked for staleness.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Refresh the shots unlocked by the runner-agent seeding and new targets:
agent runs/transcript/status, environment test pass, approval revision box,
adapter health rows, and the onboarding goal field. registry.json stamped.
Every screenshot referenced in the docs is now captured against the current
build; the only un-refreshed files are unreferenced orphans.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Close the remaining gap so every screenshot referenced in the docs is current.
seed.mjs — process-adapter "runner" agents (no LLM needed; the process adapter
executes a shell command):
- "Task Runner" runs `sh -c "echo ..."` → a real completed run with a STDOUT
transcript (run-transcript-view, test-environment-success via the config
Test button which now passes: command sh is executable)
- "Live Task" runs `sh -c "sleep 3600"` and is woken last, so it stays
in-flight through the whole capture phase → genuine "running" agent status
and an in-progress run row
routes.mjs — targets for run-transcript-view, run-history-in-progress,
agent-status-running, test-environment-success (click "Test" → wait for the
Passed result), approvals/revision-request-input (the detail note box),
adapters/health (the adapter rows), and onboarding/goal-field (wizard step 1).
dashboard.md — the dashboard image pointed at a hand-annotated PNG that can't
be auto-captured; repoint it to the freshly captured dashboard-overview.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Full automated recapture via the extended pipeline against a richly-seeded
demo instance: 220 screenshots (110 targets x light/dark, 1440x900 @2x).
Adds coverage for dashboard, tasks, activity, org chart, routines, skills,
workspaces, approvals, budget-utilisation states, new-agent forms, and
empty-state onboarding. registry.json route/depends_on backfilled and
captured_sha stamped for all 220 entries.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Grow capture coverage from 56 to 110 targets so the docs' UI screenshots are
refreshed end to end, not just the plain-URL surfaces.
capture.mjs — interaction engine:
- `steps` per target (click / fill / waitFor / press) run after the page
settles; each step is best-effort (logs + continues) so selector drift
degrades to a partial capture instead of aborting
- `clip` captures a single element (e.g. a dashboard panel), falling back to
full-page if the element isn't found
- robust locator DSL: testid / role+name / label / title / placeholder / text / css
routes.mjs — +54 targets across dashboard, tasks, issues, activity, org,
routines, skills, workspaces, settings, costs (incl. budget-bar 50/80/100%
states), new-agent forms, approvals, and empty-state onboarding. New seed-id
tokens for empty company, budget-state agents, http agent, skill, approvals.
Documents the 8 shots that are genuinely not auto-capturable offline (live-run
transcripts/status, real adapter test-pass, a nonexistent revision input, a
nonexistent adapter-health route, the hand-annotated overview, onboarding goal).
seed.mjs — extended dataset: a second empty company for empty-state shots;
three budget-utilisation agents seeded to 50/80/100% (last one paused); an
http-adapter agent; four approval items (hire / strategy / board / approved)
via REST; a resolved skill id; and the isolated-workspaces experimental flag
(PATCH /api/instance/settings/experimental).
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Aron Prins