Task

Implement Data Ingestion Module

created 3/19/2026, 10:50:49 AM • updated 3/19/2026, 10:50:49 AM

project: Gold Trading Analysis

Adaptive Task Framework

Expected Output

Decision-grade research brief with findings, evidence, tradeoffs, and recommendation

Review Owner

Destination Surface

Task-linked Brief or Campaign Assets

Fallback If Blocked

If blocked, attach the best partial artifact plus blocker note and create/trigger one recovery objective

researchbuildreview

Research

active

Requirements, constraints, failure mode, dependencies, and existing system behavior

Done when

Evidence-backed findings and a recommendation are attached.

Bound artifacts

No research artifact bound yet.

Build

pending

Implement the actual code, system change, fix, or configuration with visible output

Done when

Working implementation or systems change with verification notes and known gaps is attached with verification notes.

Bound artifacts

No build artifact bound yet.

Review

pending

Check correctness, verification quality, regressions, and readiness to ship

Done when

A clear pass/fail decision with exact fixes or approval conditions exists.

Bound artifacts

No review artifact bound yet.

New and updated tasks automatically carry the research → build → review framework, and this task now stores structured phase state so sub-agents inherit a consistent quality bar.

Title

Description

Implement the module responsible for ingesting data from identified sources into the sentiment analysis pipeline.
Workflow Contract:
EXPECTED OUTPUT: Decision-grade research brief with findings, evidence, tradeoffs, and recommendation
DESTINATION SURFACE: Task-linked Brief or Campaign Assets
REVIEW OWNER: TC
FALLBACK IF BLOCKED: If blocked, attach the best partial artifact plus blocker note and create/trigger one recovery objective
Adaptive Three-Phase Framework:
PHASE 1 — RESEARCH
Objective: Claims, counterclaims, sources, contradictions, and decision tradeoffs.
Required output: Evidence-backed findings, key constraints, and the best recommendation for the task.
Sub-agent standard: Research runs must return sources/evidence, uncertainty notes, and a recommendation.
PHASE 2 — BUILD
Objective: Turn the findings into a brief someone can make a decision from immediately.
Required output: Decision-grade research brief with findings, evidence, tradeoffs, and recommendation.
Sub-agent standard: Build runs must return the actual artifact/change plus verification notes.
PHASE 3 — REVIEW
Objective: Check source quality, uncertainty handling, and whether the recommendation is decision-grade.
Required output: Clear pass/fail decision, exact fixes if not ready, and approval conditions if partially ready.
Sub-agent standard: Review runs must judge readiness independently instead of redoing the whole task vaguely.
Cross-Phase Delegation Rule:
- Any spawned sub-agent should stay inside one phase unless the requester explicitly asks for multi-phase work.
- If a sub-agent crosses phases, it must separate its output by phase so quality can be audited cleanly.
Definition of Done:
- Complete the work through Research, Build, and Review unless the task clearly does not need one of those phases
- Produce the named output
- Link or attach the artifact where stated
- Add EVIDENCE and VERIFICATION before REVIEW/DONE
- Keep scope tight and reviewable
EVIDENCE: (fill)
VERIFICATION: (fill)

Status

Priority

Owner

Tasks now carry an adaptive framework by default. Next: phase-aware UI controls and artifact-linked completion checks.