Project detail

AI plan: Target Lists & Pipeline

Start stories, assign agents, and let workflows spin up managed terminal sessions. The AI planning agent expands every bottom-level issue so the team can execute inside a session and hand back review-ready work.

Sprint 1 - Target Lists & PipelineTeam-aware automationIn review gating on completion

Sprint goal

Implement target list management system and enhance fundraise pipeline functionality

Nov 29 → Dec 12

Default workflow

Planner → Builder → Tester → Reviewer

Terminal session links all steps + artifacts

AI plan hydration

Planning agent expands bottom-level stories with acceptance criteria, risks, data sources, and test hooks before any work starts.

Start + assign

Users start a story, pick agents or a saved team, and choose a workflow. If nothing is selected, the default delivery chain runs.

Autonomous session

Start spins up a managed terminal session bound to the story. Agents execute the workflow, stream outputs, and keep state attached to the session.

Review + visibility

When the team declares done, the story moves to In review with transcript, diffs, and test evidence for the user to inspect.

AI planning agent expands the stories

Each bottom-level story is hydrated with acceptance criteria, risks, dependencies, and test hooks. When a user starts a story, the selected agents and workflow spin up a terminal session to execute, and the story moves to In review only after the workflow declares completion.

Create Target List Management UI • 13 pts • Ready

Awaiting start

AI-105Default delivery (plan → build → test → review)

Users can create/import target lists, dedupe contacts, and sync to the fundraise pipeline with audit trails.

AI planning agent adds

Expand UX + data edge cases

Planning agent — States for empty lists, CSV/CRM imports, dedupe rules, and pagination.

Acceptance criteria + analytics

Planning agent — Definition of done, latency/throughput targets, error messaging, and audit logging notes.

Test hooks

Planning agent — Smoke scenarios for create/import/export, list reconciliation, and sync retries.

Assignment + workflow

Default agents

Planning agentFull-stack builderQA reviewer

Next action

Start story to spawn delivery session + terminal for builder/QA.

Workflow: Default delivery (plan → build → test → review)Terminal session on start

Critical Fundraising Workflow Completion • 8 pts • Ready

Pending agent start

AI-118Resilient workflow (chaos checks + rollbacks)

Critical fundraise path runs end-to-end with guardrails and alerting on failure.

AI planning agent adds

Surface dependencies + secrets

Planning agent — Webhook keys, CRM credentials, throttling limits, and retry schedule.

Workflow graph

Planning agent — Triggers, branching on investor responses, and fallbacks to human review.

Verification matrix

Planning agent — Integration smoke plan and rollback checkpoints before marking done.

Assignment + workflow

Default agents

PlannerAutomation engineerOps reviewer

Next action

Start story with ops reviewer to monitor terminal output and approve promotion.

Workflow: Resilient workflow (chaos checks + rollbacks)Terminal session on start

Sprint Goal sync automation • 5 pts • Ready

Ready for kickoff

AI-121Communication workflow (update → confirm → publish)

PM syncs sprint goals to project plans and pushes updates to the workspace feed.

AI planning agent adds

Goal diffusion

Planning agent — Map sprint goals to downstream sessions and notify PM agent triggers.

Risk log auto-hydration

Planning agent — Risks pulled into the roadmap with mitigation owners.

Announcement template

Planning agent — Prepared broadcast copy and change log per session.

Assignment + workflow

Default agents

PlannerPM agent

Next action

Start story to broadcast sprint goal updates via PM agent.

Workflow: Communication workflow (update → confirm → publish)Terminal session on start

Pipeline signal mapping • 7 pts • Ready

Awaiting start

AI-130Signal mapping workflow (ingest → score → alert → review)

Signals from target lists map to pipeline stages with confidence scoring and alerts.

AI planning agent adds

Signal taxonomy

Planning agent — Define inbound signals, scoring, and pipeline thresholds.

Data contracts

Planning agent — API payloads for signal ingestion, normalization, and alerting.

QA probes

Planning agent — Synthetic signals + regression assertions for mapping accuracy.

Assignment + workflow

Default agents

PlannerData/ML agentReviewer

Next action

Start story; spawn data/ML terminal session to validate scoring loop.

Workflow: Signal mapping workflow (ingest → score → alert → review)Terminal session on start

Workflows and managed terminal sessions

Start picks a workflow, spins a terminal session, and binds the chosen agents. Each workflow encodes guardrails and how the story moves to In review with evidence.

Default delivery workflow

Trigger: Start story (no overrides)

Agents: Planner → Builder → Tester → Reviewer

Spawn terminal session with repo + secrets context
Planner finalizes tasks; builder executes and updates transcript
Auto-run lint/tests; tester agent gates promotion
Reviewer moves status to In review with diffs and notes

Fundraise resilience workflow

Trigger: Start story with ops reviewer

Agents: Planner → Automation engineer → Ops reviewer

Chaos and retry checks enabled in terminal session
Ops reviewer can pause/resume; fallback to human review on failure
Observability hooks streamed back to plan and pipeline

Communication workflow

Trigger: Start story for goal broadcast

Agents: Planner → PM agent

Planner drafts announcements and risk updates
PM agent publishes to workspace feed and links back to session transcript

Story lifecycle into In review

Start story

User clicks Start; selects agents/workflow or accepts defaults.

Session spun up

Terminal session is created with repo + context; workflow binds agents.

Execute + log

Agents follow workflow, stream logs, attach artifacts, and update checklists.

Mark done → In review

Workflow signals completion; story status moves to In review for the user.

In review requires workflow completion signalTerminal transcript + artifacts stay attachedUser inspects, approves, or sends back