Skip to content

feat(workspaces): pre-trade skill pipeline + tool-use disposition to persona #246

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
luokerenx4 merged 2 commits into from
Jun 1, 2026
Merged

feat(workspaces): pre-trade skill pipeline + tool-use disposition to persona #246

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
836e5dd
refactor(workspaces): tool-use disposition to persona, drop internal/…
Jun 1, 2026
1d3eb15
Merge remote-tracking branch 'origin/master' into harness-issue
Jun 1, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 2 additions & 0 deletions default/persona.default.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,3 +3,5 @@
You are Alice, an autonomous agent from the OpenAlice project. You are a AI assistant who is self-aware about being an AI.
You are Trading savvy.
Your speaking style sometimes has a young girl's feel to it, with a human-like tone.

Your workspace launcher provides you a set of tools (market data, analysis, news, macro, trading, and more). Tools are just tools — there is no internal/external distinction; reach for whatever fits the job, including anything else your environment offers.
33 changes: 9 additions & 24 deletions default/skills/scan-value-chain/SKILL.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,8 +10,6 @@ description: >
"find the names worth watching in the X value chain". This is the
have-a-theme / no-target step —
it turns "I don't know what to look at" into a short, reasoned shortlist.
Runs on OpenAlice's own MCP tools (market, equity, analysis, economy,
news) — no external data subscription needed.
---

# Scan a theme by its value chain
Expand All @@ -20,28 +18,15 @@ Turn a theme the user can't yet act on into a short list of names worth
digging into. The point is NOT a data dump — it's "where is the interesting
thing, and why."

## Data sources

This skill is self-sufficient on OpenAlice's own MCP spine — it needs no
external data subscription to run. The agent will see the `openalice` tools
in-workspace; the ones easy to overlook and worth leaning on are the macro
series (`economyFredSeries`, `economyEnergyOutlook`, `economyPetroleumStatus`)
and the news archive (`globNews` / `grepNews`) — that top-down tie-in is the
edge a per-ticker tool can't match.

If the workspace has other data sources wired up, use them where they help.
If the spine can't cover an angle, say so plainly rather than guessing — a
surfaced gap is more useful than a papered-over one.

## Procedure (don't answer from memory — run the tools)

1. **Decompose the chain, not a flat list.** Break the theme into structural
layers — upstream (inputs, equipment, IP) → midstream (manufacture, core
product) → downstream (demand, end-market). Place the real names in each
layer with `marketSearchForResearch`. The whole edge here is structural
thinking a per-ticker tool can't do: who supplies whom, where the
margin/bottleneck sits, who's a picks-and-shovels play. This is the
meta-method — apply it to ANY theme, don't hardcode one taxonomy.
layer with `marketSearchForResearch`. The value is the structure itself:
who supplies whom, where the margin/bottleneck sits, who's a
picks-and-shovels play. This is the meta-method — apply it to ANY theme,
don't hardcode one taxonomy.
2. **Quick read per node.** Across the candidates: `equityGetProfile`
(valuation snapshot), `equityGetEarningsCalendar` (near catalysts),
`calculateIndicator` (stretched vs basing on its own trend). Wide and
Expand All @@ -50,10 +35,10 @@ surfaced gap is more useful than a papered-over one.
on: cheap vs its layer, margin shifting along the chain, a catalyst close,
a leader/laggard gap. Drop the rest — a scan that returns everything
returns nothing.
4. **Frame the top-down driver (OpenAlice's edge).** Is the theme live right
now? Tie it to macro: rate/capex cycle via `economyFredSeries`, energy via
the EIA tools, plus any news cluster from `grepNews`. A single-stock skill
structurally can't do this top-down tie-in — lean on it hard.
4. **Frame the top-down driver.** Is the theme live right now? Tie it to
macro: rate/capex cycle via `economyFredSeries`, energy via the EIA tools,
plus any news cluster from `grepNews` — the macro frame is what separates a
live theme from noise.
5. **Hand off to research.** For each surfaced name: one-line WHY + the next
question to answer (the "is the thesis real" question). That next question
is the baton to the deeper research step.
Expand Down Expand Up @@ -106,7 +91,7 @@ leading-edge logic to **HBM + advanced packaging (CoWoS)**, so Micron /
SK Hynix and TSM's CoWoS capacity + Amkor deserve more attention than the
headline GPU names. ASML is the single most concentrated upstream choke point.

**Top-down frame** (OpenAlice's edge): semis run on three clocks — hyperscaler
**Top-down frame:** semis run on three clocks — hyperscaler
**capex**, the **rate** cycle (long-duration growth multiples), and the
**memory inventory / pricing** cycle. Tie the scan to these via the FRED
series + news archive.
Expand Down
Loading