The Stories Bundle

A team of agents that ships content

The raw material

Your team's story is buried in git logs, sessions, and metrics.

The Stories bundle turns that raw data into polished content — slides, docs, spreadsheets, and posts — across every format an Amplifier team needs.

And it does it with a roster of specialists, not a single tool.

Proof — a shipped roster

Eleven specialist agents ship in the bundle.

Per the README roster: storyteller plus ten specialists, each reachable through the stories: namespace — a shipped team, not an idea.

So why a whole team instead of one capable writer?

11

Agents per README roster (storyteller + 10 specialists)

stories:

Namespace agents are reached through

4

Automated recipes shipped with the bundle

5

Output formats supported

The complication

No single generalist writer is great at every format at once.

Every output format is its own manual workflow. Slides, docs, spreadsheets, and posts each demand different craft, structure, and audience judgment.

A roster of specialists exists precisely to answer that.

How the team works

Each agent owns one role and one audience.

Instead of one writer faking every voice, agents are organized by role and audience — technical, marketing, executive, community, data, research, strategy, adaptation.

And each of those roles is backed by real depth.

technical-writer

Deep technical docs & architecture guides

marketing-writer

External / community communication

executive-briefer

Summaries, ROI, decision-maker content

community-manager

Engagement, celebrating user wins

data-analyst

Metrics into dashboards & insights

story-researcher

Mines git repos, sessions, ecosystem

content-strategist

Story selection & planning

content-adapter

Adapts content between formats

Depth, not wrappers

These agents are deep — thousands of lines of instruction back the roster.

The README-listed agents total roughly 2,645 lines of instruction today. Specialization here has real substance, not thin prompt wrappers.

A deep roster still needs a repeatable way to run.

~2,645

Lines across the 11 README-listed agents (today)

12

Agent .md files in agents/ (adds storyteller2)

The pipeline

Four recipes and five formats wrap the roster into a repeatable pipeline.

Automation recipes drive the agents; five output formats carry the result. This is the machinery that just produced the slides in front of you.

Which sets up the reveal.

4 automation recipes

blog-post-generator, git-tag-to-changelog, session-to-case-study, weekly-digest
5 output formats

HTML PowerPoint Excel Word PDF

The reveal

This deck is the proof — it is itself Stories output.

story-researcher mined the repo; stories:storyteller built the slides you're reading right now. The bundle's value is the artifact in front of you.

The same machinery works on your data, too.

# The pipeline that built this deck
stories:story-researcher  # mined the repo
stories:storyteller       # built these slides

# Repo: ramparte/amplifier-stories
# bundle.md name: stories  v3.0.0

Status: Active

What you keep

Point the Stories team at your own raw data — it ships your content the same way.

Across audiences and formats, the pattern that produced this overview applies to your work: raw data in, polished content out.

A team of specialist agents, reachable through the stories: namespace.

Sources

Sources & Research Methodology

Repository: ramparte/amplifier-stories (branch master), checked out at /home/ramparte/dev/ANext/amplifier-stories. bundle.md name: stories, version 3.0.0.

Data-collection snapshot: 2026-02-19 (commit 7d8c184) — the point-in-time for the roster, commit, contributor, and deck-count figures. Latest commit today: 2026-07-21 (9230211).

Feature status: Active.

Commands run:

git remote -v — confirmed origin ramparte/amplifier-stories
grep -in 'agents' README.md ; cat behaviors/stories.yaml ; ls agents/*.md | wc -l — 11 per README roster / 12 .md files (adds storyteller2)
git rev-list --count 7d8c184 → 172 commits
git log 7d8c184 --format='%aN' | sort -u | wc -l → 16 contributors
wc -l agents/<11 README files> → 2,645 lines (2,796 across all 12)
ls recipes/*.yaml → 4 recipes; README/bundle.md → 5 content formats

Gaps & qualifiers: The deck's earlier "89 decks" does not reproduce as a literal command — the docs/ .html count at the 2026-02-19 tree is 90, including non-deck index/category pages. "37 days" is 36 by strict date subtraction (2026-01-14 → 2026-02-19, inclusive counting). "2,614 agent lines" does not reproduce; the current figure is ~2,645. These are kept out of the deck's asserted claims.

Primary contributors (at 2026-02-19): Sam Schillace 49/172 (~28.5%), Michael J. Jabbour 47/172 (~27.3%), across 16 contributors total.