DottaandPaperclip
3bbc2104b6
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work. > - The feature map records product entry points and verification recipes. > - The paired agent access map records APIs and permission boundaries for the same capability IDs. > - The new index in #16987 made several missing capabilities easier to find. > - Those gaps hid existing assistant, privacy, cost, and operator workflows. > - This pull request adds 15 capabilities with source references, test scope, and manual checks. > - Readers can find these workflows without treating documentation or structural checks as live proof. ## Linked Issues or Issue Description Refs: #16987. Related map work: #16025 and #16214. A search of related pull requests found no duplicate of these additions. **Issue type** Missing documentation. **Where is the issue?** `maps/feature-map/` and `maps/agent-access-map/`. **What's wrong?** The maps omit public assistant MCP connections, private task and project sharing, subscription fees, cost-recording recovery, agent API keys, keyboard entry points, and several operator CLI workflows. **Suggested fix** Add product recipes and matching access records. Show the full capability IDs in the README index and each page. Keep the existing ID format and all 175 existing IDs. Exclude AI account pools. ## What Changed - Add the `assistant-connections` family with setup, consent, device login, client use, and revocation. - Add `tasks/privacy`, `projects/privacy`, `budgets-costs/subscriptions`, and `budgets-costs/accounting-recovery`. - Add `agents/api-keys`, `agents/managed-agent-setup`, and `navigation-preferences/command-palette`. - Add `cli-operations/feedback-export`, `chat-channels/outbound-email`, and `instance-operations/runtime-setup`. - Pair all 15 additions with access rules, source evidence, manual checks, and explicit coverage gaps. Update the visible index to 35 families and 190 capability IDs. - Add cross-links from existing access, connection setup, search, and CLI recipes. Do not change the schema or add AI account pools. - Clarify that the task Share sheet displays direct and inherited access. Manage inherited access at its ancestor or project source. ## Verification - `node scripts/check-maps.mjs` passes: 35 areas and 190 paired capability IDs, with valid references and contract fields. - `node --test scripts/check-maps.test.mjs scripts/storybook-feature-index.test.mjs` passes: 17 tests. - `node scripts/storybook-feature-index.mjs` and `git diff --check` pass. - A comparison with the base confirms all 175 existing IDs remain and exactly the intended 15 IDs were added. The README index matches the discovered feature metadata. - `node scripts/storybook-catalog.mjs` cannot run because this checkout lacks `storybook/internal/csf-tools` and installed workspace dependencies. - `pnpm -r typecheck`, `pnpm test:run`, and `pnpm build` were attempted. All stopped before their task ran: the pnpm 9.15.4 bootstrap could not fetch and verify its registry packages in this environment. - Source review covers the documented UI, CLI, routes, services, and existing tests. No live browser, provider, or authorization journey was run. New access records remain `baseline`, `source-only`, and `not-run`. - Greptile and the Paperclip reviewer both rate current commit `bb8eb797b0a1dc26256faf2f0e57cbe7abc9adcf` 5/5. The inherited-access review thread is resolved. - [GitHub CI](https://github.com/paperclipai/paperclip/actions/runs/39466772379) passes for that commit, including typecheck, build, unit tests, Runner checks, and browser tests. Browser shard 7 passed after one retry for a runner shutdown. All PR checks are successful, neutral, or skipped. ## Risks - Low runtime risk: this change only edits maps. No application behavior, API, dependency, or schema changes. - Source-reviewed recipes can drift or miss deployment-specific behavior. The records state their limits and require live checks before a capability is marked verified. > 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, model ID `gpt-6.1-sol`, 475,000-token context window, xhigh reasoning effort, with tool use and code 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 searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [x] All Paperclip CI gates are green - [x] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge --------- Co-authored-by: Paperclip <noreply@paperclip.ing>
7.3 KiB
Projects and repositories — agent access
Product map: Projects and repositories.
Feature ID: projects.
Review date: 2026-10-12. Source commit: f981fa3cc10cb8663a87dec60345fbff46d54d0a.
This is a source-reviewed baseline of representative operations and known gaps. Each capability needs operation-level review before its entire scope can be qualified. No live agent execution is claimed. Follow the map contract for format, permission checks, evidence and maintenance.
lifecycle
Capability ID: projects/lifecycle.
Review: baseline.
API: present — representative operation IDs:
POST /api/companies/{companyId}/projects.PATCH /api/projects/{id}.DELETE /api/projects/{id}.
Agent access: available — Native create_project/list_projects; REST / CLI / MCP.
Policy: Company/project visibility, protected configuration/runtime policies; some repository/privacy management requires board.
Evidence: source-only — implementation at
the reviewed commit above.
Verification: not-run — exercise a listed permitted operation with actual agent
credentials/run context; inspect its result and applicable persistence/attribution.
Assert the relevant denied actor, owner, company or mode from Policy. Review the
remaining operations before qualifying the whole capability.
Gaps: Actual data operations exist; project discovery and private-resource permissions still apply.
task-context
Capability ID: projects/task-context.
Review: baseline.
API: present — representative operation IDs:
GET /api/companies/{companyId}/issues.POST /api/companies/{companyId}/issues.
Agent access: available — Native create_task/search_tasks; REST / CLI / MCP.
Policy: Company/project visibility, protected configuration/runtime policies; some repository/privacy management requires board.
Evidence: source-only — implementation at
the reviewed commit above.
Verification: not-run — exercise a listed permitted operation with actual agent
credentials/run context; inspect its result and applicable persistence/attribution.
Assert the relevant denied actor, owner, company or mode from Policy. Review the
remaining operations before qualifying the whole capability.
Gaps: Use explicit project context; task creation, assignment, and execution remain separate checks.
repositories
Capability ID: projects/repositories.
Review: baseline.
API: present — representative operation IDs:
GET /api/companies/{companyId}/project-repositories.PUT /api/projects/{id}/repositories.
Agent access: partial — Native list_project_repositories/create_project for initial
selection; board-only replacement API.
Policy: Company/project visibility, protected configuration/runtime policies; some repository/privacy management requires board.
Evidence: source-only — implementation at
the reviewed commit above.
Verification: not-run — exercise a listed permitted operation with actual agent
credentials/run context; inspect its result and applicable persistence/attribution.
Assert the relevant denied actor, owner, company or mode from Policy. Review the
remaining operations before qualifying the whole capability.
Gaps: Authorized initial selection works. Replacement of repositories on an existing project hard-requires board and needs a delegation decision.
defaults
Capability ID: projects/defaults.
Review: baseline.
API: present — representative operation IDs:
PATCH /api/projects/{id}.GET /api/projects/{id}/workspaces.
Agent access: partial — REST / native call_api for permitted fields; CLI
project/workspace.
Policy: Company/project visibility, protected configuration/runtime policies; some repository/privacy management requires board.
Evidence: source-only — implementation at
the reviewed commit above.
Verification: not-run — exercise a listed permitted operation with actual agent
credentials/run context; inspect its result and applicable persistence/attribution.
Assert the relevant denied actor, owner, company or mode from Policy. Review the
remaining operations before qualifying the whole capability.
Gaps: Description/goal/target selection exist; host workspace command mutation is denied to agents and budget policy is board-managed.
Repository attachment also writes a durable skill import binding in its transaction. Agent-created bindings derive their refresh person from authenticated responsible-user context. Project selection does not grant runtime credentials or override private project skill access. Verify automatic import and denial with the actual agent identity.
Workspace creation, update and deletion require project:write_workspace through the shared authorization evaluator in addition to project read access. Repository attachment does not bypass skill-import policy for its background discovery.
privacy
Capability ID: projects/privacy.
Review: baseline.
API: present — POST /api/companies/{companyId}/projects; PATCH /api/projects/{id}; GET and POST /api/projects/{id}/access-members; DELETE /api/projects/{id}/access-members/{memberId}. Creation and authorized updates accept visibility; use stable UUID paths for evidence.
Agent access: partial — authorized agent REST or admitted native call_api can create a private project, and eligible project/member reads retain their normal API paths. Existing project visibility and audience mutations require board authority; no dedicated agent audience-management tool is established.
Policy: Company access and project:read apply; inaccessible projects return not found. Listing members requires project read, unlike task-grant listing. Visibility/member mutations require a board privacy/personal owner, active company owner/admin, instance admin, or local implicit board. The server derives the creation owner from the board person or agent’s onBehalfOfUserId. Targets must be company agents or active company users; the owner cannot be removed. Agent access intersects agent and responsible-user access. Opening a project does not publish its protected tasks.
Evidence: source-only — reviewed at 9ef434fe79c57d976b8b03b14c5242db270b8320. Project properties, access members, project routes, and authorization.
Verification: not-run — Verify allowed creation and reads as the actual operator/agent, then compare project and task access after adding/removing a person. Reject ordinary members’ audience changes, agent member/visibility mutations, foreign-company targets, and owner removal. After opening the project, inspect protected tasks and independent direct grants rather than inferring task publication from project discovery.
Gaps: The dedicated create_project tool does not advertise a visibility field; explicit private selection uses REST or admitted call_api. Private-project creation through each client, inherited task audiences, and responsible-user combinations need separate live evidence. The map does not infer filesystem isolation from these app-record permissions.