RUNAIVA / PolicyGraph
Capture Intelligence
Mission → question → Executive Brief.
The solicitation is the track. Backend intelligence stays rich; Executive Brief stays calm and decision-oriented.
Select Capture Mission
Choose a case study or create a capture mission to begin.
RUNAIVA / PolicyGraph / Capture Intelligence
Step 1 — Pick the Mission
Mission: —
Use the mission list to switch workspaces, or create a new capture mission.
Case: DHS FLETC Student Support
A workspace can be switched at any time.
Capture Intelligence Briefing Audio
Context, not evidenceNotebookLM scene-setting audio for the DHS FLETC Student Support case study.
Use this short briefing to orient the case before reviewing Phil evidence outputs.
Contextual briefing only. This NotebookLM audio is not source evidence unless its underlying sources are separately ingested and cited.
Open NotebookLM Audio BriefingStep 2 — Load Skill
Active Skill: Capture Intelligence SkillCapture Intelligence skill active.
Mission Cockpit
Cockpit = lane receipts · Report = BLUF · Analyst = evidence · Runner = steer · Evidence references = audit trail
Executive Capture Brief
Loading deterministic lane audit…
Skill activation · telemetry
Activated vs suppressed capture lanes for the selected skill lens (inventory + intent). Timeline row labels align when Mission001 evidence supports them.
Load a mission to hydrate telemetry activation.
Evidence / Telemetry Detail
Lane receipts and MCP routing appear on each cockpit card; use Executive Capture Brief above and lane audit for deterministic proof.
—
Cockpit expanded
Ask Runner
Select a mission to activate Ask Runner.
Step 3 — Ask Runner
Active Skill: Capture Intelligence Skill
Mission selected: —
Ask from the mission evidence. Phil guides the line; you stay in control.
Phil understood your question as…
Confirm or adjust the focus before Runner starts research.
Advanced — compound intent & provenance
Runner research
Runner selected primary artifacts and activated structured evidence
Step 3 — Runner uses the active skill
Runner uses visible source cards as primary artifacts, then activates structured solicitation knowledge when available: Section C requirements, Section L instructions, Section M factors, L/M crosswalks, campaign memory, and legal/protest context.
If you already have a document, add it first.
Approved Working Set — 0 approved sources
Pinned Sources
Phil Strategy
Open when you want the current capture logic behind the next move.
Phil Input Pack
Steering guides focus and ordering only; it cannot override evidence rules or create unsupported findings.
Telemetry / Pit Wall
Market Intelligence / USAspending Signals
Price Realism / Spend Alignment
Spend intelligence supports price realism and execution-risk inference. It is not solicitation scoring and does not replace Section M.
Advanced diagnostics · Runner / CMCS telemetry
Evidence Readiness: Moderate
Advanced / Analyst & Proof Tools
Evidence workspace
Source controls
Catalog search
Filter the local evidence catalog (F1 view).
Add Data
Phil suggests. Evidence rules weigh. User can override source inclusion by adding local records manually.
USAspending (local evidence)
Search the ingested USAspending DuckDB first. Phil does not call USAspending live during analysis; public URLs are pointers only until files are ingested locally.
No local USAspending record found. Download the award data from USAspending.gov,
place it in /home/ocasey/mvp/fpds_app/Incoming/USSpend data Download, then run ingestion.
Use USAspending Advanced Search, filter by PIID / vendor / agency / FY, click Download, and save the file into the USSpend data Download folder.
Legal Evidence (COFC / GAO)
Search the local legal evidence DuckDB first. Phil does not call COFC or GAO live during analysis; public URLs are pointers only until PDFs or pages are ingested and archived locally.
No local legal record found. Download the COFC or GAO record into the Legal Evidence input folder,
then run python3 scripts/ingest_legal_evidence_directory.py.
Default input locations (override with LEGAL_EVIDENCE_INPUT_DIRS):
/home/ocasey/mvp/fpds_app/Incoming/Federal Court of Appeals and
/home/ocasey/mvp/fpds_app/Incoming/GAO Protest.
Mission Evidence Readiness
Loading…Phil will use this capture mission inventory before generating insight.
This reflects available data, not active analysis scope.
Inventory summary
Grouped counts only — full catalog is not listed here.
Phil Working Set
Phil will ONLY use the sources selected below.
Capture mission inventory shows what is available. The Working Set is what Phil will analyze.
0 selected (0 used in last run)
Review what Phil will read, or change the set.
Full source inventory
Working set cards, structured nodes, and dismissed sources.
Dismissed sources
Sources you removed from Phil’s recommendation review (Advanced).
Review Selected Evidence
Uses the capture mission inventory above.
No sources selected for this run.
Step 4 — Choose Report Type
Choose what Phil should prepare from the selected evidence.
Active Capture Mission: —
Runner selected primary artifacts and activated structured evidence
Runner used the checked source cards as primary artifacts and activated structured mission evidence internally when available.
0 of 0 selected for this output
Engineer View: selected source controls
Phil will analyze the checked primary artifacts plus activated structured mission evidence when available. Uncheck source cards to keep them in the mission but exclude them from this run.
Ask Runner to use the active skill and telemetry plan before opening report choices.
More report types
Generated Report Preview
Active Skill: Capture Intelligence Skill
Review this draft before deciding whether to hand it to Studio.
Runner Retrieval Workspace
Step 5 — Optional Studio Handoff NotebookLM demo
Open only when you want to move this draft into NotebookLM for demo output options.
Studio output will be saved here.
Use the copied report as the source for the demo notebook.
Prompt steering
Add to Capture Intelligence SME Insight
SME insight is doctrine, not proof. Runner uses these as reconnaissance lanes, Phil uses them to explain what a pattern means, but CMCS still requires governed evidence for any final claim.
Phil Feedback Loop
Tell Phil what worked, what missed, and how to improve the next answer.
Feedback saved for Phil.
Recent feedback
Phil Input Pack
Send to Studio
Use the generated RÚN Briefing to prepare a downstream deliverable.
Add Source
Browse disk on Sources, or pin a placeholder.
New Capture Mission
Name this workspace. You can switch capture missions anytime from the list.