Intel Brief — Analyst to Writer relay
researchfeatured
Installation
$
npx agenticloops install intel-briefSummary
Two teammates in one loop: an analyst extracts what changed, hands off, and a writer turns it into a one-paragraph team brief.
- Runs Every 4h, on any harness.
- Relay: analyst (dude) → writer (theo). Each role hands its output to the next via
{{previous_output}}— one run, one signed receipt. - Runs at the frontier capability tier (smartest model); the harness picks the model.
- Proof: the reference multi-agent loop for SPEC §4; every run signs a chain receipt
What it does
An analyst extracts the few shifts that matter from the last interval and returns three bullets. A writer receives them via {{previous_output}} and turns them into a one-paragraph team brief. One scheduled loop, two roles, one structured handoff.
LOOP.md
--- name: intel-brief description: Intel Brief — Analyst to Writer relay · Two teammates in one loop: an analyst extracts what changed, hands off, and a writer turns it into a one-paragraph team brief. schedule: every 4h tier: frontier # frontier | standard | fast effort: medium agents: # ordered chain of roles · one run · structured handoff - role: analyst persona: dude prompt: | From the last interval's activity, pick the 3 shifts most worth the team's attention. Return only 3 short bullets. - role: writer persona: theo prompt: | # receives {{previous_output}} from the prior role Turn the analyst's 3 bullets into a single tight paragraph (max 60 words), lead with what matters most. tags: [research, market-intel, multi-agent, relay] license: MIT ---