Files
W1---Opportunity-Intelligence/docs/SPRINT0.md
T

3.5 KiB

W1 — Opportunity Intelligence / Sprint 0

Historical operating record. Commands and root paths below describe the original layout; use the current README and migration notes for executable procedures.

W1-016 offline qualification: see QUALIFICATION.md for the reproducible process, audit rules and generated opportunity brief.

Authorized kickoff: 2026-09-17. Research only. No applications, outreach, purchasing, private accounts, or production deployment.

First cycle

opportunities.json contains five distinct direct-source opportunities discovered on September 17. id tracks pipeline identity; source_id is the original requisition/solicitation key. Do not mistake the 5 entries for five qualified offers, confirmed open positions, or revenue. Three recruiter requisitions and two procurement RFPs. No separately confirmed salaried-only opening in this batch. Links and rationale must be rechecked before external action.

Collection proof and manual fallback

LRS automated discovery is now available separately: see LRS.md for commands, public-index access, source defects, evidence fields and output files. Run python lrs_discovery.py and test both sources with python -m unittest discover_test discover_dates_test lrs_test -v. The older Sprint 0 limitations below describe the initial collection only. Run python3 discovery.py on a networked machine with Python 3, from this folder. It collects public USAC procurement links into runs/YYYY-MM-DD/usac.json, compares solicitation identifiers with opportunities.json, and writes new_candidates.json. It does not infer opening status from presence on the index. If blocked, open https://www.usac.org/about/procurement/ manually, inspect the live index, visit the individual solicitation, and enter it in opportunities.json with original URL and checked date. For jobs, manually inspect the three direct recruiter URLs in the seed pipeline; no job-board scraping is claimed automated.

Run python3 -m unittest discover_test.py for fixture-based extraction/dedup regression. Retain dated snapshots. Do not scrape sites requiring login or bypass access restrictions. A scheduled invocation and full three-category unattended collection are NOT installed yet.

Minimum record

id, category, organization, title, source_id, url, posted, checked, location, engagement, compensation, status, rationale, next_action; deadline when applicable. Explicit not verified rather than fabricated values. Dedup by source organization plus source_id, then manually inspect semantic duplicates and related recruiter engagements.

Review

Ken checks originals, chooses pursue / hold / exclude, verifies mandatory requirements and start/payment timing, and separately authorizes applications, proposals or outreach. Report counts by category, changes since previous run, source errors, time/spend, and exceptions. Never treat job pay or RFP estimated value as actual income.

Known limitations

Only one public procurement index is programmatically collected; the first search and five seed records were researched interactively. The script may fail on source-layout or network changes; procurement index contains closed items; no unattended schedule, broad job feed, mail connector, notification or CRM integration is proven. First cycle has no verified salaried-only or standalone part-time posting. Actual owner time must be entered by Ken; no incremental spend has been initiated by this artifact.