Core concepts at a glance
Every concept Disco Parrot uses on one page, one sentence each, each one a click to its full guide. The map of the documentation.
This page is the map. Every concept Disco Parrot uses, defined in a sentence, with a link to its full guide. Read it top to bottom for the bird's-eye, or skim for the names so you know what to click on next. For a strict A-to-Z lookup instead, the glossary defines every term in a line.
The work model
The records that hold your team's work and the relationships between them.
- Projects. The container a team's work lives in. Holds initiatives, binds to a repository, sets the default sandbox profile.
- Portfolios. The rollup container above projects. Team-based membership, with a privacy switch. For leaders looking across several projects at once.
- Initiatives. The captured intent. Clarifications, Revise Spec, dependsOn graph, and the spec body that decomposes into plans.
- Plans. The reviewable pieces an initiative decomposes into. Seven plan types cover implementation, design, spec, review, research, chore, and verification.
- Bugs. Issues that live alongside plans. They roll up against initiatives and can be assigned to a sprint.
- Sprints. Team-scoped overlays that group plans and bugs into a delivery window.
- Goals and OKRs. Outcomes the work supports. Key results tie back to initiatives that contribute to them.
The SDLC work model page walks through these in order, with the relationships drawn out.
How the agent works
The objects that turn intent into action.
- Flows. Orchestrations of agent work. A flow has ordered steps, parameters, checkpoints, and the option to run interactively or in the background.
- Skills. Named, reusable instructions for one kind of work. A skill can run on its own from the action launcher, or as a step inside a flow.
- Sessions. The unified view of every chat, flow run, and background task in your workspace, with their full transcripts.
- Human checkpoints. The moments a flow pauses for approval. You Approve, Reject, or Skip and the run continues from there.
What shapes the agent
The configuration that decides how the agent behaves on your team's work.
- Agents and Agent Instructions. An "agent" is the runtime plus the instructions you write. Agent Instructions are versioned and scoped: global or per sandbox profile.
- AI models. The runtimes you enable: Claude, Codex, Google Gemini, GitHub Copilot. Pick a tenant default; users and skills can override.
- Status-scoped skills. Bindings that decide which skills appear on which records at which statuses, so the right action is offered at the right moment.
Where the work runs
The execution surface.
- Sandboxes. Disposable containers, one per run. The agent and a sidecar run inside; all I/O goes through the sidecar. The container is the boundary.
- Approved actions. The three mechanisms that bound the agent's reach: tool allowlists on MCP servers, propose-only environment policies, and credentials the agent never holds.
How change is governed
The records that keep the platform honest.
- Workflows. The status machines a record moves through. Five workflows ship (initiative, plan, bug, flow, test-case), each with its own statuses and transitions you can customize. Distinct from Flows, which are agent orchestrations.
- Audit trails. Three complementary records. The tenant audit log, the per-execution Sessions transcript, and per-entity version history. Together they answer who did what, what the agent did inside the run, and what changed.
- Entity versioning and history. Non-destructive snapshots on every initiative, plan, skill, and agent instruction with an AI badge and a restore that creates a new version rather than overwriting.
How the data is structured
The way records are stored, addressed, and queried.
- Built for query. Every record is structured data with a stable URL. The same records that you filter, sort, and share by URL are the records the agent reads at chat time and writes back at the end of the turn.
- Repository documentation. The wiki the platform keeps for each repository, generated by AI, kept fresh by documentation health reviews you approve.
How it connects
The external systems the platform talks to.
- Integrations and providers. Work-tracking integrations (Azure DevOps, GitHub), code-repo providers (GitHub, GitHub Enterprise), document providers (Google Drive, OneDrive), and notification channels (in-app, email, Microsoft Teams).
Working in Disco Parrot
The surfaces you actually click on day to day. These are task pages, not concepts, but they are how you do the work.
- List views and the data grid. The grid every entity gets, with column filters, the filter panel, inline cell editing, bulk edit with select-all-matching, and shareable URL-encoded views.
- Boards (Kanban). The same records arranged by status. The sprint board is the first mixed-entity board with plans and bugs in the same columns.
- Roadmap and timelines. Initiatives drawn on a time axis with drag-to-reschedule, edge-resize, dependency arrows, and sub-task rollups.
- Quick View. Peek any record in a modal over your list without losing your place.
- Command Center. The live panel of every chat, flow run, and background task you have in flight.
- Action launcher. The drawer that starts a manual skill, a chat, or a flow from any record.
- Navigation, workspaces, and search. The three areas, the side rail, the project switcher, the URL conventions.