Change runner from runners-farhoodlabs to ubuntu-latest across all fork
workflows. Update container registry from ghcr.io to git.farh.net and
authenticate with REGISTRY_TOKEN. Migrate update-infra API calls from
GitHub to Gitea. Disable refresh-lockfile.yml (requires GitHub gh CLI).
Update CLAUDE.md references.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Closes the coverage gap on the actual migrated function. Mocks the
two network-touching git-source exports (resolveGitRef, openRepoSnapshot)
while keeping parseGitSourceUrl real so the parseGitHubSourceUrl shim
contract stays honest. Adds 5 cases:
- happy path: opens one snapshot, calls listFiles, readFileOptional
on COMPANY.md, readFile on candidate paths
- ref fallback: when openRepoSnapshot('main') rejects, falls back to
'master' and emits the expected warning
- COMPANY.md absent everywhere: throws "missing COMPANY.md"
- referenced logo: readBinary is called for the logoPath from
.paperclip.yaml
- logo read failure: warning emitted, no throw
57/57 portability tests passing; existing 52 unchanged via shim.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Mirrors the skills refactor: company-portability was the second user of
the per-host REST shim (its own parallel parseGitHubSourceUrl + fetch
helpers + raw.githubusercontent URL builder), so importing a company
package from a non-github URL hit the same Gitea 404 the skills path did.
- Extend git-source.ts:
- parseGitSourceUrl: also recognises query-string shape
(?ref=...&path=...) used by portability URLs, with precedence over
path-style segments when both are present.
- RepoSnapshot: add readBinary (Uint8Array for the company logo
fetch) and readFileOptional (null on NotFoundError, for the
COMPANY.md probe + main->master fallback).
- Rewrite resolveSource in company-portability.ts to open a single
in-memory snapshot per import and serve all reads (COMPANY.md,
candidate tree, includes, logo) from it. Drops fetchText/fetchJson/
fetchBinary/fetchOptionalText.
- parseGitHubSourceUrl stays exported with its original return shape
({hostname, owner, repo, ref, basePath, companyPath}) so the existing
test suite passes unchanged. It now delegates URL parsing to
parseGitSourceUrl and layers companyPath derivation on top.
- Delete server/src/services/github-fetch.ts: zero remaining callers.
Test coverage:
- 7 new git-source tests (query-string parse variants, query-string
precedence over path style, readBinary, readFileOptional NotFound
null + non-NotFound rethrow) — 34/34 passing.
- 52 existing company-portability tests still pass via the
parseGitHubSourceUrl shim contract.
- Smoke-tested end-to-end against https://git.farh.net/.../?ref=main:
ref resolves, snapshot opens, readFile/readBinary/readFileOptional
all return expected results.
Note: two pre-existing failures in company-skills-routes.test.ts
("does not expose a skill reference...") exist on dev too and are
unrelated to this change.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- Resolves the duplicate-SHA conflict on the gitea/skills commits
by taking dev's versions (canonical after PR #13 superseded the
original shim with the git-source refactor).
- Deletes .github/workflows/assemble-local.yml -- the workflow
triggered on master push but lived on local, so it never fired
automatically; promotion happens via dev->local PRs instead.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
27 tests covering the surface that had none:
- parseGitSourceUrl: bare URLs (github/gitea/gitlab), tree/blob/src
shapes, subpaths, file paths, trailing .git stripping, https-only
enforcement, malformed/missing-segment rejection.
- resolveGitRef: 40-hex SHA passthrough (no network call), default
branch via HEAD symref, named branch, peeled annotated tag, lightweight
tag, ref-not-found, network/401/404 error translation, onAuth
callback shape (token-as-username, x-oauth-basic) and absence.
- openRepoSnapshot: clone args (singleBranch/depth=1/noCheckout),
tree walk filtering trees vs blobs, readFile path, SHA fallback
when tracking ref is null, 404 translation.
Mocks at the isomorphic-git boundary; verifies our adaptation logic,
not isomorphic-git itself.
Known limit surfaced by a test (not fixed here): gitea URLs with
slash-containing branch names like /src/branch/feature/x are
ambiguous without server-side disambiguation. The test uses a
single-segment branch; the multi-segment case needs a separate fix
(refCandidates from longest-to-shortest, resolved against
listServerRefs output).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The skill import/update/file-read pipeline talked to host-specific REST
APIs (GitHub /commits/{ref}, /git/trees/{sha}, raw.githubusercontent.com)
and the recent Gitea support was a parallel shim on top of the same
pattern. The result was multiple ref-resolution shapes that needed
per-host branching, and on Gitea the /commits/{ref} endpoint returns
404 outright -- so even public Gitea/Forgejo repos failed to import.
Replace with a single git-source module backed by isomorphic-git +
memfs. It speaks the smart-HTTP protocol any sane git server already
serves:
- resolveGitRef: one listServerRefs call, no host API. Handles default
branch (symref on HEAD), named branches, annotated/lightweight tags,
and SHA passthrough.
- openRepoSnapshot: shallow singleBranch clone into an in-memory fs;
listFiles via git.walk, readFile via git.readBlob. No tempdirs, no
execFile, no per-host endpoints.
- Universal auth via onAuth (token-as-username) covering GitHub PATs,
GitLab PATs, Gitea/Forgejo tokens.
- parseGitSourceUrl recognises github tree/blob, gitea src/branch|
commit|tag, gitlab /-/tree, bitbucket /src/{ref} URL shapes plus
bare clone URLs.
Stored skill metadata is unchanged (hostname/owner/repo/ref/trackingRef/
repoSkillDir), so existing rows keep working -- the clone URL is
derived at fetch time.
company-portability.ts still imports github-fetch.ts (same broken
pattern, separate feature). Left as a follow-up.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The skills source pipeline was hardcoded to GitHub conventions, so even
though the UI now accepts non-GitHub URLs, the server couldn't actually
fetch from anywhere else.
- github-fetch.ts: dispatch by host family (github.com → GitHub API +
raw.githubusercontent.com; everything else → Gitea/Forgejo API v1 +
/api/v1/repos/.../media for raw content).
- parseGitHubSourceUrl: also accept Gitea/Forgejo web URLs
(/{owner}/{repo}/src/{branch|commit|tag}/{ref}/{path}).
- routes/company-skills.ts: drop the hostname='github.com' gate in
deriveTrackedSkillRef so non-GitHub skills are still tracked.
- Generalize user-facing strings ('GitHub PAT' → 'PAT', 'GitHub source URL'
→ 'Source URL', etc.).
GitHub Enterprise (was assumed by '/api/v3') is no longer a special case —
non-github.com hosts are treated as Gitea/Forgejo. If GHE support is needed
later, add a per-source host-family override.
The skills source pipeline was hardcoded to GitHub conventions, so even
though the UI now accepts non-GitHub URLs, the server couldn't actually
fetch from anywhere else.
- github-fetch.ts: dispatch by host family (github.com → GitHub API +
raw.githubusercontent.com; everything else → Gitea/Forgejo API v1 +
/api/v1/repos/.../media for raw content).
- parseGitHubSourceUrl: also accept Gitea/Forgejo web URLs
(/{owner}/{repo}/src/{branch|commit|tag}/{ref}/{path}).
- routes/company-skills.ts: drop the hostname='github.com' gate in
deriveTrackedSkillRef so non-GitHub skills are still tracked.
- Generalize user-facing strings ('GitHub PAT' → 'PAT', 'GitHub source URL'
→ 'Source URL', etc.).
GitHub Enterprise (was assumed by '/api/v3') is no longer a special case —
non-github.com hosts are treated as Gitea/Forgejo. If GHE support is needed
later, add a per-source host-family override.
The PAT input on the skill import flow was hidden by a regex that matched
github.com or org/repo shorthand. Self-hosted Gitea/Forgejo/GitLab sources
got no auth field at all. Always show the input when a source is entered,
and label it generically ('Personal access token') instead of 'GitHub PAT'.
UI only — backend already accepts any token via /skills/:id/auth and
/companies/:companyId/skills POST {source, authToken}.
The PAT input on the skill import flow was hidden by a regex that matched
github.com or org/repo shorthand. Self-hosted Gitea/Forgejo/GitLab sources
got no auth field at all. Always show the input when a source is entered,
and label it generically ('Personal access token') instead of 'GitHub PAT'.
UI only — backend already accepts any token via /skills/:id/auth and
/companies/:companyId/skills POST {source, authToken}.
Adds the official Gitea 'tea' CLI (v0.14.0) alongside the existing forgejo
CLIs (fj, fj-ex, fgj). Useful when interacting with Gitea instances whose API
surface is covered by tea but not by the forgejo variants.
Adds the official Gitea 'tea' CLI (v0.14.0) alongside the existing forgejo
CLIs (fj, fj-ex, fgj). Useful when interacting with Gitea instances whose API
surface is covered by tea but not by the forgejo variants.
Adds focused tests for every code path the agentId addition touches:
- environment-runtime.test.ts (4 new tests):
- plugin-driver acquireLease forwards agentId in RPC payload when present
- plugin-driver acquireLease omits agentId from RPC payload when null
- sandbox-provider acquireLease forwards agentId when present
- sandbox-provider resumeLease forwards agentId when reuseLease=true matches
- seedEnvironment helper now exposes the seeded agentId
- environment-run-orchestrator.test.ts (2 new tests):
- acquireForRun threads agentId through to runtime.acquireRunLease
- logActivity records the same agentId on environment.lease_acquired
- new vi.hoisted mocks for environmentService.getById + ensureLocalEnvironment
- agent-test-environment-routes.test.ts (1 new assertion):
- ad-hoc operator test-environment probe calls acquireRunLease with
agentId: null and heartbeatRunId: null (no agent context)
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
## Summary
- Add `releases/v2026.513.0.md` covering the stable release range
`v2026.512.0..origin/master` (6 PRs).
- Includes one new DB migration (`0084_issue_recovery_actions`) under
the Upgrade Guide.
- No breaking changes detected; all PRs are core-maintainer commits so
the Contributors section is omitted.
## Highlights captured
- Source-scoped recovery actions
([#5599](https://github.com/paperclipai/paperclip/pull/5599))
- Blocked Inbox attention view
([#5603](https://github.com/paperclipai/paperclip/pull/5603))
- Local plugin development workflow
([#5821](https://github.com/paperclipai/paperclip/pull/5821))
## Test plan
- [ ] Reviewer confirms the highlight/improvement/fix categorization
matches release intent
- [ ] Reviewer confirms `0084_issue_recovery_actions` upgrade note is
accurate
- [ ] Reviewer signs off on `releases/v2026.513.0.md` for the stable
release cut
Generated under [PAP-9378](/PAP/issues/PAP-9378) via the
`release-changelog` skill.
Co-authored-by: Paperclip <noreply@paperclip.ing>
## Thinking Path
> - Paperclip orchestrates AI agents for zero-human companies through
company-scoped issues, comments, approvals, and execution workspaces.
> - Operators need the Inbox to show not only active work, but also
blocked work that may need human or agent attention.
> - The existing inbox experience did not have a dedicated blocked-work
surface, so blocked tasks were harder to triage and resume deliberately.
> - Backend consumers also needed a compact attention signal that
distinguishes actionable blockers from covered or waiting blocker
states.
> - This pull request adds a Blocked Inbox tab backed by issue
blocker-attention metadata, shared validators, and UI helpers.
> - The benefit is a clearer triage path for stalled or blocked
Paperclip work without exposing external wait internals in the
operator-facing UI.
## What Changed
- Added shared issue blocker-attention types, validators, and exports
for the API/UI contract.
- Added backend blocker-attention computation and issue route support
for blocked inbox data.
- Added the Blocked Inbox tab, blocked reason chips, filtering/search
UI, responsive layouts, and Storybook stories.
- Updated inbox helpers and page behavior so toolbar controls only
appear where they apply.
- Added coverage for shared validators, server blocker-attention
behavior, blocked inbox UI helpers/components, and the Inbox page.
- Added a screenshot helper script for the blocked inbox Storybook
stories.
- Addressed Greptile feedback by making urgency sorting deterministic
for null stop times, avoiding full blocked-inbox list enrichment for
counts, and hardening the screenshot helper.
## Verification
- Rebased the branch cleanly onto `public-gh/master`.
- Confirmed the diff does not include `pnpm-lock.yaml`.
- Confirmed the diff does not include database migration files.
- Ran `pnpm exec vitest run packages/shared/src/validators/issue.test.ts
server/src/__tests__/issue-blocker-attention.test.ts
ui/src/components/BlockedInboxView.test.tsx
ui/src/components/BlockedReasonChip.test.tsx
ui/src/lib/blockedInbox.test.ts ui/src/lib/inbox.test.ts
ui/src/pages/Inbox.test.tsx`.
- Ran `pnpm --filter @paperclipai/shared typecheck && pnpm --filter
@paperclipai/server typecheck && pnpm --filter @paperclipai/ui
typecheck`.
- Checked `ROADMAP.md`; this is scoped inbox/operator triage work and
does not duplicate a listed roadmap feature.
- Greptile Review is green on the latest head and all four Greptile
review threads are resolved.
- GitHub PR checks are green on the latest head: policy, security/snyk,
e2e, verify, Canary Dry Run, Greptile Review, and serialized server
suites 1/4 through 4/4.
## Risks
- Medium review surface because this touches the shared issue contract,
server issue services, and the Inbox UI together.
- Blocker-attention classification may need product tuning after
operators use it on real blocked queues.
- UI screenshots were not attached in this PR-opening pass; the branch
includes `scripts/screenshot-blocked-inbox.mjs` and Storybook stories
for visual capture.
> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected — check the roadmap
first. See `CONTRIBUTING.md`.
## Model Used
OpenAI Codex, GPT-5-based coding agent with shell, git, GitHub CLI,
GitHub connector, and Paperclip API tool use. Reasoning mode: medium.
Context window: not exposed by the runtime.
## Checklist
- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [ ] If this change affects the UI, I have included before/after
screenshots
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge
---------
Co-authored-by: Paperclip <noreply@paperclip.ing>
## Thinking Path
> - Paperclip orchestrates AI agents through a control plane while
adapters execute work in local, remote, or sandboxed runtimes.
> - Remote sandbox execution depends on a strict host-versus-remote
workspace boundary: the host prepares/restores files, while the adapter
command runs inside the sandbox cwd.
> - Jannes' PR #5823 identified host-side failure modes that were not
covered by replacement PR #5822.
> - Persisting a remote pod cwd in session params could poison the next
host heartbeat resume and make Paperclip inspect or upload system temp
roots.
> - Plugin sandbox providers also need a narrow way to receive
model-provider API keys without exposing the full server environment to
every plugin worker.
> - This pull request ports the host-side fixes from #5823 in the
current codebase style, with focused regression coverage.
> - The benefit is safer remote sandbox resumes and plugin worker
environment handling without broadening core plugin privileges.
## What Changed
- Persist host workspace cwd, not remote sandbox cwd, in `claude_local`
session params while retaining remote execution identity metadata.
- Reject saved session cwds that point at system roots before heartbeat
falls back to agent home workspace.
- Skip sockets, FIFOs, devices, and other non-file entries during
workspace restore snapshot capture/comparison.
- Pass a small model-provider API-key allowlist only to plugins
declaring `environment.drivers.register`.
- Added focused regression tests for remote Claude session params,
unsafe session cwd detection, plugin worker env filtering, and non-file
snapshot entries.
Credits: ports host-side fixes from Jannes' #5823.
## Verification
- `pnpm vitest run
packages/adapter-utils/src/workspace-restore-merge.test.ts
server/src/services/session-workspace-cwd.test.ts
server/src/__tests__/claude-local-execute.test.ts
server/src/__tests__/plugin-database.test.ts` (25 passed, 7 skipped by
existing embedded-Postgres host guard)
- `pnpm --filter @paperclipai/adapter-utils typecheck`
- `pnpm --filter @paperclipai/adapter-claude-local typecheck`
- `pnpm --filter @paperclipai/server typecheck`
## Risks
- Low risk: changes are scoped to remote sandbox/session metadata,
workspace snapshot filtering, and plugin worker env setup.
- Sandbox-provider plugins now receive only the explicit model-provider
key allowlist; any provider needing another key name will need a
deliberate allowlist update.
> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected — check the roadmap
first. See `CONTRIBUTING.md`.
## Model Used
- OpenAI Codex, GPT-5-based coding agent, tool-enabled local code
execution and repository editing.
## Checklist
- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] If this change affects the UI, I have included before/after
screenshots
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge
---------
Co-authored-by: Paperclip <noreply@paperclip.ing>
## Thinking Path
> - Paperclip orchestrates AI-agent companies through company-scoped
issues, comments, and execution context.
> - The issue detail page is the board surface where operators and
agents inspect a task in its parent/child workflow.
> - Ordered sub-issues need a low-friction way to move through work
without returning to the parent list after every issue.
> - Existing issue detail navigation only covered sibling transitions
and did not continue into a parent issue's first ordered child.
> - This pull request adds ordered previous/next navigation for issue
detail views and extends it to continue from a parent or last sibling
into the first direct child.
> - The benefit is a smoother review/execution path through hierarchical
work while preserving hidden issue filtering and dependency-aware
ordering.
## What Changed
- Added `IssueSiblingNavigation` and route-state handling so issue
detail footers can link to previous/next ordered issues.
- Extended sub-issue ordering helpers to build navigation from siblings
plus direct children, including root-parent and
last-sibling-to-first-child cases.
- Added page, component, and library tests for ordered sibling
navigation, child fallback navigation, hidden issues, and link
rendering.
- Fixed the quicklook blur/click race Greptile found by deferring close
until after portaled link clicks can complete, with a regression test.
- Polished the navigation landmark label so it remains accurate when the
next target is a direct child rather than a sibling.
## Verification
- `pnpm exec vitest run src/components/IssueLinkQuicklook.test.tsx
src/lib/issue-detail-subissues.test.ts
src/components/IssueSiblingNavigation.test.tsx
src/pages/IssueDetail.test.tsx --config vitest.config.ts` from `ui/` -
31 tests passed.
- `pnpm --filter @paperclipai/ui typecheck` - passed.
- `git diff --check` - passed.
- GitHub PR checks on latest head `34046be2` - passed: Greptile Review,
verify, e2e, Canary Dry Run, policy, Snyk, and serialized server shards.
- Screenshots: not captured in this heartbeat; this PR is a draft and
the changed states are covered by focused component/page tests.
## Risks
- Low risk; this is a UI navigation addition with no database or API
contract changes.
- The main behavioral risk is navigation ordering drift if
`workflowSort` expectations change later.
- The IssueDetail navigation now waits for child issue loading, which
avoids stale child fallback links but can delay footer navigation
briefly while data loads.
> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected - check the roadmap
first. See `CONTRIBUTING.md`.
## Model Used
- OpenAI Codex, GPT-5 coding agent with repository tool use and shell
execution.
## Checklist
- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [ ] If this change affects the UI, I have included before/after
screenshots
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge
---------
Co-authored-by: Paperclip <noreply@paperclip.ing>
## Thinking Path
> - Paperclip is the control plane for autonomous AI companies, and
issue comments are the primary durable communication surface between
operators and agents.
> - Commit `c445e592` (`fix(ui): fix message attribution for
agent-posted comments with user author IDs (#5780)`) added server-side
derived attribution for historical comments by scanning heartbeat runs
near comment timestamps.
> - That scan accidentally bound JavaScript `Date` objects directly into
postgres-js SQL fragments for the run timestamp window.
> - On real Postgres, that can fail while listing issue comments with
`ERR_INVALID_ARG_TYPE`, which makes comments disappear from issue pages
such as `PAP-9284`.
> - This pull request keeps the attribution behavior intact while
changing only the broken timestamp binding path.
> - The benefit is that comments load again without weakening the
conservative attribution recovery introduced by `c445e592`.
## What Changed
- Convert the derived-attribution heartbeat-run window bounds to ISO
timestamp strings before binding them into SQL, with explicit
`::timestamptz` casts.
- Add an embedded Postgres regression that inserts a heartbeat run and
user-authored comment, then verifies `issueService.listComments()`
returns the comment while the attribution scan runs.
- Delete `heartbeat_runs` during the issue service test cleanup before
deleting agents so the new test data does not leak across cases.
## Verification
- `pnpm exec vitest run server/src/__tests__/issues-service.test.ts -t
"lists user comments when derived run attribution scans a timestamp
window"`
- `pnpm --filter @paperclipai/server typecheck`
- `git diff --check`
## Risks
- Low risk. The change is limited to how timestamp parameters are bound
for an existing query.
- The derived attribution logic remains conservative and still requires
exact run-log proof before relabeling a comment.
- The regression uses embedded Postgres so it covers the postgres-js
binding path that failed in production-like local runs.
> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected — check the roadmap
first. See `CONTRIBUTING.md`.
## Model Used
- OpenAI Codex via the Paperclip `codex_local` adapter; GPT-5
coding-agent family with local terminal, file-editing, and git/GitHub
CLI tool use. Exact hosted model deployment ID is not exposed by this
local adapter runtime.
## Checklist
- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] If this change affects the UI, I have included before/after
screenshots (not applicable: server-side comment API bugfix)
- [x] I have updated relevant documentation to reflect my changes (not
applicable: no documented behavior or command changed)
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge
Co-authored-by: Paperclip <noreply@paperclip.ing>
## Thinking Path
> - Paperclip is the control plane for autonomous AI-agent companies.
> - Plugins are the extension point for adding capabilities without
expanding the core product surface.
> - Local plugin development needed a tighter CLI-first loop so plugin
authors can scaffold, run, install, inspect, and reload plugins without
reaching into internal package paths.
> - The server plugin install path also needed local-path handling that
keeps plugin identity, dashboard routes, and development watchers
coherent.
> - This pull request adds the CLI scaffold/install workflow, fixes the
server and SDK edge cases that blocked that loop, and updates the
agent-facing plugin creation skill and docs.
> - The benefit is that contributors can develop plugins from local
folders with a documented, repeatable happy path.
## What Changed
- Added `paperclipai plugin init` coverage and CLI wiring for local
plugin scaffolding.
- Improved local plugin install handling, plugin key route resolution,
dashboard capability behavior, and dev watcher startup/reload behavior.
- Fixed plugin SDK worker entrypoint validation for symlinked package
layouts.
- Added targeted tests for plugin init, server plugin authz/watcher
behavior, SDK worker host validation, and the authoring smoke example.
- Added a short local plugin development guide and refreshed the plugin
authoring guide plus `paperclip-create-plugin` skill instructions.
## Verification
- `pnpm run preflight:workspace-links && pnpm --filter
@paperclipai/plugin-sdk build && pnpm --filter
@paperclipai/create-paperclip-plugin typecheck && pnpm --filter
paperclipai typecheck && pnpm --filter @paperclipai/plugin-sdk typecheck
&& pnpm --filter @paperclipai/server typecheck`
- `pnpm exec vitest run --project paperclipai
cli/src/__tests__/plugin-init.test.ts`
- `pnpm exec vitest run --project @paperclipai/plugin-sdk
packages/plugins/sdk/tests/worker-rpc-host.test.ts`
- `pnpm exec vitest run --project @paperclipai/server
server/src/__tests__/plugin-dev-watcher.test.ts --pool=forks
--poolOptions.forks.isolate=true`
- `pnpm exec vitest run --project @paperclipai/server
server/src/__tests__/plugin-routes-authz.test.ts --pool=forks
--poolOptions.forks.isolate=true`
- `pnpm --dir packages/plugins/examples/plugin-authoring-smoke-example
test`
- Confirmed `pnpm-lock.yaml` is not included in the PR diff.
## Risks
- Medium risk: this touches plugin install routing, CLI command
behavior, and the local development watcher.
- Local path plugin installs execute trusted local code by design; the
new docs call out that trust boundary.
- No database migrations are included.
> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected — check the roadmap
first. See `CONTRIBUTING.md`.
## Model Used
- OpenAI Codex, GPT-5 coding agent, tool-enabled local shell and git
workflow, medium reasoning effort. Context window details were not
exposed in this runtime.
## Checklist
- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] If this change affects the UI, I have included before/after
screenshots
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge
UI screenshots: not applicable; this PR changes CLI/server/plugin docs
and tests, not board UI rendering.
---------
Co-authored-by: Paperclip <noreply@paperclip.ing>
## Thinking Path
> - Paperclip is a control plane for autonomous AI companies, where work
must end with a clear disposition rather than ambiguous agent liveness.
> - Recovery currently detects stalled or missing-next-step issues, but
source issue recovery can become split across child recovery issues,
blockers, and comments.
> - That makes it harder for operators and agents to see who owns
recovery and what exact action is needed on the original issue.
> - Source-scoped recovery actions give the original issue a first-class
active recovery state with owner, evidence, wake policy, and resolution
outcome.
> - This pull request adds the recovery-action data model, backend
reconciliation and resolution APIs, and board UI indicators/actions.
> - The benefit is clearer stalled-work recovery without losing source
issue context or relying on comments as the liveness path.
## What Changed
- Added the `issue_recovery_actions` schema, shared
types/constants/validators, and an idempotent
`0084_issue_recovery_actions` migration ordered after current `master`
migrations.
- Updated stranded/missing-disposition recovery to create source-scoped
recovery actions, wake the recovery owner on the source issue, and avoid
locking the source issue for recovery-action wakes.
- Added API support for reading active recovery actions on issue
detail/list surfaces and resolving them with restored, blocked,
cancelled, or false-positive outcomes.
- Require blocked recovery resolutions to have an unresolved first-class
blocker, and removed the UI shortcut that could mark recovery blocked
without a blocker selection path.
- Surfaced recovery indicators/actions in the issue UI, blocker notices,
active run panels, issue rows, and Storybook coverage.
- Updated docs and focused tests for recovery semantics, ownership,
races, stale comments, and UI behavior.
## Verification
- `pnpm exec vitest run
server/src/__tests__/issue-recovery-actions.test.ts
server/src/__tests__/heartbeat-process-recovery.test.ts
ui/src/components/IssueRecoveryActionCard.test.tsx
ui/src/components/IssueBlockedNotice.test.tsx ui/src/api/issues.test.ts`
— 5 files, 72 tests passed.
- `pnpm --filter @paperclipai/shared typecheck` — passed.
- `pnpm --filter @paperclipai/db typecheck` — passed, including
migration numbering check.
- `pnpm --filter @paperclipai/server typecheck` — passed.
- `pnpm --filter @paperclipai/ui typecheck` — passed.
- Follow-up verification after blocker-resolution guard: `pnpm exec
vitest run server/src/__tests__/issue-recovery-actions.test.ts
ui/src/components/IssueRecoveryActionCard.test.tsx
ui/src/api/issues.test.ts` — 3 files, 27 tests passed.
- Follow-up `pnpm --filter @paperclipai/server typecheck` — passed.
- Follow-up `pnpm --filter @paperclipai/ui typecheck` — passed.
- UI states are available in
`ui/storybook/stories/source-issue-recovery.stories.tsx`; screenshot
capture helper is `scripts/screenshot-recovery-card.cjs`.
## Risks
- Medium: recovery behavior changes from child recovery issue ownership
toward source-scoped actions, so operators may see stalled-work state in
new places.
- Migration risk is mitigated by using the next migration slot after
`master` and making the table/constraints/index creation idempotent for
anyone who previously applied the old branch-local
`0082_dizzy_master_mold` migration.
- Existing child recovery issue paths are still guarded for
already-created recovery issues, but new source-scoped flows should be
watched in CI and Greptile review.
> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected — check the roadmap
first. See `CONTRIBUTING.md`.
## Model Used
- OpenAI Codex, GPT-5 coding agent, tool use enabled for shell, Git,
GitHub, and local test execution. Context window not exposed by the
runtime.
## Checklist
- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] If this change affects the UI, I have included before/after
screenshots
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge
---------
Co-authored-by: Paperclip <noreply@paperclip.ing>
Symmetric with the acquireLease change. Lets plugin-backed sandbox
providers reject a reusable lease whose stored agentId doesn't match
the current run's agent, forcing the host to acquire a fresh lease
instead of stomping the previous agent's workspace state.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Thread agentId into PluginEnvironmentAcquireLeaseParams + host call sites
so plugin-backed sandbox providers (e.g. paperclip-plugin-k8s) can scope
lease state per-agent without needing an SDK callback or DB lookup.
Add an optional agentId field to PluginEnvironmentAcquireLeaseParams and
thread it through the host's environment-runtime + run-orchestrator call
sites so plugin-backed sandbox providers can scope lease state (subdirs,
PVCs, etc.) per agent without an SDK callback or DB lookup.
The field is required-but-nullable on the internal EnvironmentDriverAcquireInput
(string | null) so every call site has to think about whether it has an
agent context. Ad-hoc operator probes (agent test-environment route)
pass null. The plugin RPC payload omits the field entirely when null,
keeping wire compatibility with older plugin worker SDKs.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
## Thinking Path
> - Paperclip’s issue chat is an audit surface: reviewers need to trust
who actually authored a message.
> - Some historical agent comments were persisted with `authorUserId`
and no surviving `createdByRunId`, so the UI rendered real agent output
as if it came from the board user.
> - A pure timestamp-window fallback is too risky because human
reviewers can comment while agents are running.
> - The safe recovery path is to derive attribution only when the server
can prove it from same-issue run logs that include the exact posted
comment id, then let the chat renderer prefer that recovered agent
attribution.
> - This keeps historical threads trustworthy without mutating old
database rows or guessing in ambiguous cases.
## What Changed
- Added shared `IssueComment` fields for derived attribution so server
and UI can carry recovered `derivedAuthorAgentId`,
`derivedCreatedByRunId`, and `derivedAuthorSource` consistently.
- Added server-side attribution recovery in
`server/src/services/issues.ts` that reads same-issue run logs and only
derives agent authorship when a run log contains the exact `comment id:
...` emitted during posting.
- Updated issue chat rendering in `ui/src/lib/issue-chat-messages.ts` to
prefer direct agent authorship, then activity-log `runAgentId`, then the
server-derived attribution.
- Removed the unsafe UI-only run-window fallback from
`ui/src/pages/IssueDetail.tsx` so human comments posted during an active
run are not silently relabeled as agent output.
- Added regression coverage for both the run-log derivation path and the
chat-rendering fallback behavior.
- Bounded server-side run-log enrichment to 8 concurrent reads per
request and removed the unused `issueCommentSchema` declaration during
PR cleanup.
## Verification
- `pnpm exec vitest run ui/src/lib/issue-chat-messages.test.ts
server/src/__tests__/issues-service.test.ts`
- `pnpm test:run:general`
- Live validation on May 12, 2026 in `PAPA-322`: confirmed the
previously misattributed historical comments on `PAPA-316` now render as
Claude-authored on `http://goldie.gerbil-company.ts.net:3100`.
- Reviewer check: open `PAPA-316` in the running instance and confirm
historical comments such as `## Investigation: exe.dev 422 + codex
re-test` render under Claude instead of the board user.
## Risks
- Low risk. The change is scoped to comment attribution recovery and
rendering.
- Derived attribution is intentionally conservative: if there is no
exact run-log proof, the comment remains user-authored instead of
guessing.
- Run-log recovery depends on retained same-issue logs, so older
comments without that evidence remain unchanged.
## Model Used
- OpenAI Codex via the Paperclip `codex_local` adapter (GPT-5-class
coding agent with tool use in the local Paperclip runtime; the exact
deployment/model ID is not surfaced by this workspace).
## Checklist
- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [ ] If this change affects the UI, I have included before/after
screenshots
- [ ] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge
---------
Co-authored-by: Paperclip <noreply@paperclip.ing>
## Summary
PR [#5366](https://github.com/paperclipai/paperclip/pull/5366) already
merged the v2026.511.0 changelog. This follow-up bumps the artifact to
the actual cut date and drops the pre-alpha sandbox work per maintainer
feedback.
- **Rename** `releases/v2026.511.0.md` → `releases/v2026.512.0.md`
- **Bump header / date** to `# v2026.512.0` / `> Released: 2026-05-12`
- **Drop new sandbox content** (pre-alpha, not yet ready):
- Daytona sandbox provider plugin highlight
([#5580](https://github.com/paperclipai/paperclip/pull/5580),
[#5586](https://github.com/paperclipai/paperclip/pull/5586))
- Cursor sandbox support improvement
([#4803](https://github.com/paperclipai/paperclip/pull/4803))
- Cursor sandbox runtime resolution fix
([#5446](https://github.com/paperclipai/paperclip/pull/5446))
- Sandbox provider messaging polish
([#4902](https://github.com/paperclipai/paperclip/pull/4902))
- **Add LLM Wiki plugin package highlight**
([#5716](https://github.com/paperclipai/paperclip/pull/5716)) — the
package itself landed on master after #5366 merged.
- **Update Upgrade Guide closer** to mention only the `cursor_cloud`
adapter as opt-in.
The `cursor_cloud` adapter is kept in (adapter, not sandbox). The
exe.dev and Cloudflare sandbox provider plugins that landed since the
merge are also excluded as pre-alpha.
No breaking changes; the nine new migrations (`0075`–`0083`) carry over
unchanged from the merged 511 file.
## Test plan
- [ ] Maintainer review of the dropped entries — confirm I caught
everything sandbox-related you wanted out
- [ ] Confirm Cursor Cloud adapter staying in is intentional (flag for
removal if not)
- [ ] Confirm LLM Wiki plugin package highlight phrasing
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-authored-by: Paperclip <noreply@paperclip.ing>
Chris Farhood
Dotta