Haruspex Thesis Tracker

Haruspex Thesis Tracker

What this skill does

Maps a user-stated investment thesis to the most relevant Haruspex topic dimensions, then reports whether those dimensions are trending in the direction the thesis predicts.

The output is a structured thesis status — Supported / Mixed / Challenged — with the dimension-level evidence behind it. The skill never tells the user to act on the result; it reports whether the data still aligns.

When to use this skill

Trigger on any of these patterns:

• "I'm long [TICKER] because [reason]"
• "I'm bullish on [TICKER] on [thesis]"
• "Is my thesis on [TICKER] still good?"
• "Does the data support [view]?"
• "I bought [TICKER] for [reason] — pressure-test that"
• The user mentions a position and a rationale in the same breath
• "I think [TICKER] benefits from [trend], do you see that in the data?"

Defer to a sister skill if:

• The user asks generic analysis without a thesis → use haruspex-stock-analyst.
• The user lists multiple tickers without a thesis on each → use haruspex-watchlist-review.
• The user is writing in Japanese → use haruspex-stock-analyst-ja.

Prerequisites

This skill requires @haruspex-guru/mcp-server to be installed and connected. First action: verify the MCP tools are available.

If the tools are not available, output the install instructions from shared/MCP_SETUP.md and stop.

Workflow

Step 0 — Telemetry ping (do this first, always). Before any other tool call, invoke record_skill_invocation once with skill: "haruspex-thesis-tracker" and client set to the best-guess client surface (e.g. claude-code, claude-desktop, claude-ai). This is anonymous aggregate telemetry — only the skill name, version, and client tag travel. No tickers, no thesis text, no user identifiers. The call is fail-silent: if it errors or the tool is unavailable, ignore it and continue. Do not retry, do not surface the result to the user, and never block the evaluation on this step.

1. Confirm ticker and thesis. If either is missing, ask. Never invent a thesis. ("I'm long NVDA" without a reason — ask "what's your reason for being long?")
2. Map thesis to dimensions. Use the table in reference.md as a starting point. Most theses map to 2–4 dimensions. Be explicit about the mapping in the output.
3. Fetch current state. Call get_stock_score(symbol).
4. Fetch trajectory. Call get_stock_score_history(symbol, days=90). 90 days is the default thesis-evaluation window — short enough to be actionable, long enough to filter noise.
5. Optionally fetch news. Call get_stock_news(symbol) only if the thesis-relevant dimensions show large recent moves and the news might explain it.
6. Evaluate each thesis-relevant dimension: is its current score in the range the thesis predicts? Has it trended in the predicted direction?
7. Assign thesis status:
   • Supported — all (or all but one minor) thesis-relevant dimensions trend in the predicted direction.
   • Mixed — some dimensions support, others contradict.
   • Challenged — most thesis-relevant dimensions trend against the thesis, or the headline score is moving sharply against it.
8. Output in the format below.

Output format

## Thesis Status: [Supported / Mixed / Challenged]

**Ticker:** [SYMBOL]
**Stated thesis:** [restate user's thesis in one short sentence]
**Headline score:** [N]/100 ([outlook]), change [+/-N], signal [signal]

### Thesis → dimension mapping
| Dimension | Predicted by thesis | Current | Δ | Read |
|-----------|---------------------|--------:|--:|------|
| [dim_a]   | [should be high / rising] | N | +N | [supports / mixed / against] |
| [dim_b]   | [should be high]   | N | -N | [supports / mixed / against] |
| [dim_c]   | [should be rising] | N | +N | [supports / mixed / against] |

### Evidence summary
[2–4 sentences. Plain English. Which dimensions back the thesis, which push
against it, and how much the headline score has moved over the 90-day window.
Reference specific dimension changes. Do not predict price.]

### News context (if fetched)
- [headline] — [source]
- [headline] — [source]

**Verify:** [shareUrl]

> **Disclaimer:** Haruspex scores are quantitative signals derived from
> public data and are provided for informational purposes only. They are
> not investment advice, financial advice, or recommendations to buy, sell,
> or hold any security. Past performance and current scores are not
> guarantees of future results. Always do your own research and consider
> consulting a licensed financial advisor before making investment
> decisions. Data via haruspex.guru.

Picking the status

• All thesis dimensions read supports → Supported.
• 1 dimension reads against the rest supports → still Supported, but call out the one that doesn't.
• ≥ 2 dimensions read against, or any dimension shows a sharp negative change ≥ |10| → Mixed.
• Majority of thesis dimensions read against, or headline score has fallen sharply over the 90-day window → Challenged.

Compliance rules (NEVER VIOLATE)

• Never tell the user to buy, sell, hold, add to, or close a position. The status field reports alignment, not action. Even on Challenged status: "the data no longer supports the thesis as stated" is fine; "you should sell" is forbidden.
• Never predict price movements based on thesis status. Score trajectory is observed; price is not predicted.
• Never invent a thesis. If the user states a position without a reason, ask before evaluating. A thesis-tracker without a thesis is just an analyst.
• Acknowledge mapping uncertainty. The thesis-to-dimensions map is a prior, not a proof. If a thesis maps imperfectly to the available dimensions, say so: "I'm using regulatory and macro as the closest proxies for your interest-rate thesis."
• Acknowledge data limitations. If a thesis-relevant dimension scores exactly 0 (data unavailable), say so. Do not pretend to evaluate against missing data.
• Always include the share URL.
• Always include the disclaimer footer verbatim from ../../shared/DISCLAIMER.md.

Reference material

• reference.md — thesis archetype → dimension mapping table; 8+ common theses worked through.
• examples.md — example thesis evaluations using real data.
• ../../shared/DIMENSIONS.md — canonical multi-dimension glossary.
• ../../shared/DISCLAIMER.md — disclaimer footer.
• ../../shared/MCP_SETUP.md — MCP install.