investment-os
AGENTS.md1 starsChanged 6 days agoMITView on GitHub
Your job is to install and evaluate this repository without changing the user's global environment or creating an automated investment service. Investment OS is a multi-market research system. Preserve the U.S./Hong Kong daily watchlist path, market and macro proxies, and optional global research sources when adding or changing a market-specific adapter. a-share-daily is additive and stock-specific; do not make it the default for other markets, route Hong Kong or U.S. symbols through it, or remove other source families merely because…
Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.
Of the AGENTS.md files we read, the 33,315 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repositoryOne other file it gives its agents.
- investment-researchSkill
The file
We split this AGENTS.md into the rules it gives an agent
6 rules. About 1 is common, 2 are distinctive, and 3 are very specific to this project.
≈1
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- could be automatic: a setting could enforce it
- Very specific: line 12
Run uv sync --frozen --extra dev inside the checkout.
- Distinctive: line 16
Read the generated brief and report.
- Very specific: line 41
Use scripts/verify_dependency_profiles.py when the task is to verify optional package availability without calling providers or adding …
- Common: line 48
Do not add credentials to files.
- Very specific: line 66
When the user asks for an A-share observation or daily brief, run investment-os a-share-daily and return the contents of brief.md first.
- Distinctive: line 67
Keep commands, test counts, adapter names, status codes, logs, and paths out of the reader brief.
≈ “Common” is an estimate: at least 1 in 5 of the 42,283 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Agent instructions
Your job is to install and evaluate this repository without changing the user's global environment or creating an automated investment service.
Market-scope contract
Investment OS is a multi-market research system. Preserve the U.S./Hong Kong daily watchlist path, market and macro proxies, and optional global research sources when adding or changing a market-specific adapter. a-share-daily is additive and stock-specific; do not make it the default for other markets, route Hong Kong or U.S. symbols through it, or remove other source families merely because an A-share task is active.
Required path
- Confirm Python is 3.11 or 3.12 and
uvis available. - Run
uv sync --frozen --extra devinside the checkout. - Run
uv run investment-os doctor. - Run
uv run investment-os demo --out demo-output. - Run
uv run pytest -q. - Read the generated brief and report. Confirm they contain research questions and evidence limits, not trade instructions.
- Return the exact commands, pass/fail state, and output paths.
Live-source test
Live access is optional. Use the isolated market profile only when the user asks for a current-source check:
cp configs/watchlist.sample.yaml configs/watchlist.local.yaml
uv run --frozen --isolated --link-mode copy --extra market \
investment-os run --config configs/watchlist.local.yaml --out live-output --strictFor an A-share request, use the separate keyless China profile and stock-specific command:
uv run --frozen --isolated --link-mode copy --extra china \
investment-os doctor --probe china-keyless
uv run --frozen --isolated --link-mode copy --extra china \
investment-os a-share-daily \
--config configs/a-share-watchlist.sample.yaml \
--out .local/a-share-agent-test \
--strictUse scripts/verify_dependency_profiles.py when the task is to verify optional package availability without calling providers or adding those packages to .venv:
uv run --no-project python scripts/verify_dependency_profiles.py market
uv run --no-project python scripts/verify_dependency_profiles.py chinaDo not add credentials to files. Do not print credential values. Do not enable cron, notifications, brokerage access, or order execution during evaluation.
Acceptance
Evaluation passes only when:
- doctor exits
0; - offline demo exits
0and writes all five artifacts; - tests pass;
scripts/public_release_guard.pypasses;- any live run reports at least one
data_quality=okrow when--strictis used.
A report file with zero usable source rows is a failed live test, not a successful run.
Reader output contract
The acceptance record above is the engineering receipt, not the default user answer.
- When the user asks for an A-share observation or daily brief, run
investment-os a-share-dailyand return the contents ofbrief.mdfirst. - Keep commands, test counts, adapter names, status codes, logs, and paths out of the reader brief. Add at most one compact pass/fail receipt after it.
- Detailed evidence belongs in
evidence-ledger.csv,coverage-matrix.md, andsource-receipt.json; show it only when the user asks to audit or troubleshoot. no_eventis a successful source check with no matching event.source_erroris a failed source read. Never collapse one into the other.- Tushare is an optional second source for the keyless A-share path. Missing
TUSHARE_TOKENdoes not makea-share-dailydegraded. - Turnover, generic fund-flow fields, and margin financing are not verified institutional net flow. Do not rename them to imply otherwise.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

