B608: Bandit flags any f-string containing "select" as potential SQL
injection. All four cases (app.py, ds_trace.py, llm_st.py, merge.py)
are Streamlit UI labels or log messages — not database queries.
B701: Jinja2 autoescape=False warnings in coder.py and utils.py are
false positives — these render Python code and plain-text templates,
not HTML. Enabling autoescape would corrupt the rendered code.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- submit.py: eval(json_str) → ast.literal_eval(json_str) for safe
Python-literal parsing without arbitrary code execution
- info.py: add timeout=30 to both requests.get() calls to prevent
indefinite hangs on unresponsive GitHub API
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- factor.py: check_output([python_bin, path]) instead of shell string
- env.py QlibCondaEnv: all four conda commands use list args
Shell=True with a constructed string allows shell injection if
python_bin or path contain shell metacharacters.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The metric parameter was passed directly into an f-string SQL query.
Add explicit validation against _ALLOWED_METRICS before use, raising
ValueError on unknown values. Raises ValueError on injection attempt
instead of silently accepting arbitrary column names.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The previous "fix" introduced pd.read_pickle(encoding="utf-8", "/path")
which is a SyntaxError (positional argument after keyword argument).
pd.read_pickle() has no encoding parameter.
Replace with correct # nosec B301 comment — pickle is safe here because
the files are written by the Kaggle preprocessing pipeline in a sandboxed
container and never sourced from user input.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Replace f-string SQL queries with whitelist validation:
- Table name must be in _ALLOWED_TABLES
- Column name must be alphanumeric+underscore
- Column type must be in _ALLOWED_COL_TYPES
- Use pragma_table_info() for existence check instead of SELECT f-string
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Switch log rotation from midnight-only ("00:00") to size-based:
per-command logs: 50 MB, all.log: 100 MB (with gz compression)
- Shorten retention from 30/60 days to 7 days
- Cap llm_calls.jsonl entries to 500 chars per field to prevent
GB-scale files from long-running loops
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
EUR/USD synthetic data has \$volume=0 for all rows, causing any VWAP or
volume-weighted factor to produce all-NaN output. Insert a guard after
pd.read_hdf() that replaces zero volume with (\$high - \$low) range proxy
so volume-dependent factors produce meaningful signals.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
LLM sometimes copies the .reset_index(level=N, drop=True) suffix from
groupby().rolling().method() patterns and adds it after .transform(),
but transform() already preserves the original index. The extra
reset_index() drops an index level and causes ValueError: 'cannot reindex
on an axis with duplicate labels' or shape mismatch on assignment.
Detect: any line containing both .transform( and .reset_index(level=..., drop=True)
Fix: strip the .reset_index() suffix from those lines.
Adds 1 new test (test_transform_reset_index_stripped) — total 30 tests.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
1. _fix_instrument_column_access: var['instrument'] = EXPR was incorrectly
converted to var.index.get_level_values(1) = EXPR, producing a SyntaxError
('cannot assign to function call'). Added (?!\s*=) negative lookahead to
skip assignment targets.
2. _fix_groupby_column_on_multiindex: groupby(['instrument','date']) on a
reset_index() variable was converted to groupby([var.index.get_level_values...])
but reset_index() produces a plain RangeIndex, not a MultiIndex, causing
AttributeError: 'RangeIndex' has no attribute 'normalize'. Added reset_vars
guard to skip variables produced by reset_index().
Adds 1 new test (test_assignment_target_not_touched) — total 29 tests, all passing.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
1. groupby(level=['instrument','date']) → get_level_values() — string level
names like 'date' don't exist in the (datetime, instrument) MultiIndex;
replaced with get_level_values(0).normalize() + get_level_values(1).
2. groupby(level=['date','instrument']) — symmetric fix for reversed order.
3. groupby(level=['instrument']) → groupby(level=1) — single string level.
4. groupby(level=N)['col'].apply(lambda) → transform(lambda) — apply() on a
grouped Series prepends an extra index level, causing index shape mismatch
when assigned back; transform() preserves the original index.
5. df.loc[instrument] DateParseError fix (instrument_loc_multiindex) — already
committed, adding supporting tests for groupby(level=['instrument','date']).
Adds 5 new tests (TestGroupbyLevelStringNames, TestGroupbyApplyToTransform)
— total 28 tests, all passing.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
When LLM iterates over instruments via get_level_values('instrument').unique()
and then does df.loc[instrument], pandas tries to parse the instrument string
('EURUSD') as a datetime against level-0 of the (datetime, instrument) index,
raising DateParseError.
Fix: detect loop variables bound to get_level_values(1) or get_level_values('instrument')
and replace DF.loc[loop_var] (read) with DF.xs(loop_var, level=1). Assignment
write-backs are left untouched to avoid complex rewrites.
Adds 4 new tests (TestInstrumentLocMultiindex) — total 23 tests, all passing.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
LLM-generated code often accesses df['instrument'] as a column, but
'instrument' is an index level (level 1) in the MultiIndex DataFrame.
Replace with df.index.get_level_values(1) except when the variable
was created via reset_index() (where the column actually exists).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Two bugs that together caused an infinite SKIP loop after LoopResumeError:
1. loop.py _run_step: set step_forward=False in the `else: raise` branch so that
when LoopResumeError propagates from _propose (LLMUnavailableError), step_idx
stays at 0. Previously it advanced to 1, leaving loops permanently stuck with
missing direct_exp_gen result on next resume.
2. base.py _create_chat_completion_auto_continue: when finish_reason=="length"
triggers a continuation retry, merge into the previous assistant message instead
of appending a second consecutive one. llama-server returns 400 on two consecutive
assistant messages, which caused LLMUnavailableError -> LoopResumeError cascade.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
LLM generates invalid Python by putting keyword args inside lists:
df.groupby([level=1, 'date']) ← SyntaxError
Also fixes the regex for the chained groupby Pattern A/B which had
an unescaped ')' causing re.error that silently reverted the fix.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The fixer was raising min_periods to match window size, which causes
all-NaN output for intraday factors with 96 bars/day — window=240 means
zero valid bars per day, window=60 means 61% NaN per day. Critics were
consistently flagging this as incorrect for intraday factors. The LLM
now controls its own min_periods.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
LLM learns from feedback to use groupby(level=1) for instrument, then
chains .groupby('date') to add the date dimension — but DataFrameGroupBy
has no .groupby() method, causing AttributeError at runtime.
Replace the invalid chain with a correct two-level groupby using
index.get_level_values(), consistent with the existing instrument+date fix.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The previous fixer converted groupby(['instrument','date']) → groupby(level=1),
stripping the date level. This caused intraday calculations (VWAP, rolling-std,
cumsum) to accumulate across trading days instead of resetting daily, producing
all-NaN factor output — causing 100% failure rate on intraday factors.
New behaviour: capture the DataFrame variable name and emit:
var.groupby([var.index.get_level_values(1),
var.index.get_level_values(0).normalize()])
which groups by (instrument, day) as originally intended.
Adds test/qlib/test_auto_fixer.py covering all fixer cases.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The LLM generates x.rolling(window=N, ddof=1).std() where ddof is passed
to rolling() instead of std() — pandas raises TypeError on any ddof in rolling().
Fix both forms: rolling(..., ddof=N) and rolling(...).std(ddof=N).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- _fix_reset_index_groupby: replace groupby(level=N) on reset_index'd variables
with groupby('instrument') — fixes ValueError: level > 0 only valid with MultiIndex
- _fix_groupby_mixed_levels: strip string level names from groupby(level=[int, 'str'])
to fix AssertionError: Level 'date' not in index
- _fix_groupby_column_on_multiindex: convert groupby(['instrument','date']) on
MultiIndex DataFrames to groupby(level=1) — fixes KeyError on column access
- _fix_rolling_ddof: remove unsupported ddof kwarg from rolling().std()/var()
- fix(proposal): apply history compression to factor_proposal.py (was causing
131k-token prompts from QlibFactorHypothesis2Experiment; pycache had stale .pyc)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Summarize all but the 2 most recent experiments to compact bullet lines
(factor name, PASS/FAIL, IC value, 120-char observation snippet) instead
of including full verbatim traces; reduces prompt from ~121k to ~40-60k tokens
- Fix _evaluate_factor_directly and _save_factor_values to look for result.h5
and factor.py in sub_workspace_list instead of experiment_workspace
- Fix Series.to_parquet() → Series.to_frame().to_parquet() in _save_factor_values
- Update factor_data_template README: correct bars-per-day (1440, not 96)
- Update prompts to accept 2024-only debug dataset output as valid factor result
- Fix factor_coder prompts: allow 2024 debug data in date-range instruction
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Rule 7 extended: session-based aggregations (London/NY/Asian) must also
be shifted by 1 trading day before use — same as daily aggregations
- Rule 8 added: prefer pure intraday rolling factors (RSI, Bollinger, VWAP
deviation, rolling std) that have no look-ahead risk and vary every minute
- predix_full_eval.py: apply _shift_daily_constant_factor_if_needed before IC
- predix_gen_strategies_real_bt.py: improved swing prompt with daily-level
signal logic guidance for daily-constant factors
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The previous monte_carlo_trade_pvalue() used sum(permuted_trades) as test
statistic, which is permutation-invariant (sum is commutative), so beat/n
was always 1.0 and MC_p was always 1.00 for every strategy.
Replace with a one-sided binomial test on trade win rate vs 50% baseline.
Tests whether the observed win rate could occur by chance under H0: p=0.5.
Also add _shift_daily_constant_factor_if_needed() to predix_full_eval.py
so re-evaluations apply the look-ahead bias correction for daily factors.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Daily factors (e.g. daily_log_return) carried same-day close data at 00:00,
giving the model end-of-day information at bar open — a classic look-ahead bias
that produced spurious IC=0.25 and Sharpe=24 with 98% win rate.
Changes:
- factor_runner.py: add _shift_daily_constant_factor_if_needed() that detects
factors where >90% of days have a single unique intraday value, then shifts
them by 1 trading day before IC computation
- prompts.yaml: add rule #7 instructing LLM to always shift(1) daily aggregates
before forward-filling to minute bars
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- B301 (pickle): add nosec B301 to pd.read_pickle calls in Kaggle templates
— files are trusted Kaggle-environment inputs, not user-supplied
- B614 (torch.load): add weights_only=True to all torch.load calls in
model benchmark GT code and gt_code.py
- B104 (binding 0.0.0.0): change run_server and CLI default to 127.0.0.1;
add nosec comment where all-interface binding is required for Docker
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- monte_carlo_trade_pvalue(): shuffles trade P&L N times, returns fraction of
permuted sequences that beat real total return (p<0.05 = genuine edge)
- walk_forward_rolling(): multiple IS/OOS windows (IS=3yr, OOS=1yr, step=1yr),
computes wf_oos_sharpe_mean, wf_oos_consistency (% profitable windows)
- backtest_signal_riskmgmt(): new wf_rolling and mc_n_permutations params
- Strategy generator: enables both (200 MC permutations), adds mc_ok and wf_ok
to acceptance filter (mc_p<0.20, wf_consistency>=50%)
- Rebacktest script: enables both, stores all wf_*/mc_* fields in write-back
- 6 new tests covering MC pvalue, disabled-by-default, zero-trades edge case,
rolling WF key presence and consistency range
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Split IS (2020-2023) and OOS (2024-2026) periods with independent RiskMgmt
simulations. Strategy acceptance now requires OOS sharpe > 0 and
OOS monthly return > 0 to prevent overfitting. OOS metrics stored in
strategy JSON summary and CSV reports.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Replace sequential predict() calls with predict_batch() in both
build_kronos_factor and evaluate_kronos_model. Up to batch_size windows
processed simultaneously on GPU, reducing per-window time from ~10s to
~0.13s (10 windows in 1.3s on RTX 5060 Ti, 75x speedup).
Adds --batch-size / -b option (default 32) to both kronos-factor and
kronos-eval CLI commands. Falls back to single inference per window if a
batch fails. Refactors timestamp prep into _build_window_inputs helper.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
KronosPredictor.predict() requires x_timestamp and y_timestamp to be
pandas Series of datetime values for its calc_time_stamps() helper.
Previously we passed integer ranges (after reset_index), which raised
AttributeError on .dt.minute. Fixed by extracting datetime index values
before resetting and using future_idx for y_timestamp.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Move top-level `import torch` into _cuda_available() helper so
kronos_adapter.py can be imported in CI environments without torch.
All device defaults resolved at runtime via lazy detection.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Add vbt_backtest.py as single source of truth for all metric formulas
(Sharpe, drawdown, IC, transaction costs) — backtest_engine.py and
strategy_orchestrator.py now delegate to it
- Add LLMUnavailableError to exception.py; rd_loop.py catches it at the
proposal stage and raises LoopResumeError to avoid corrupting trace
history with None hypotheses
- Guard record() against None exp/hypothesis so loop resets leave
trace.hist in a consistent state
- Refactor strategy_orchestrator and optuna_optimizer to use unified
backtest path; remove duplicate metric calculation code
- Add predix_rebacktest_unified.py script for offline re-evaluation
- Update tests and README
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Path injection (#22, #28, #29, #30):
- Switch from Path.relative_to() to os.path.realpath() + str.startswith()
in all four path-validation sites across finetune and rl UI data_loader.py
and finetune app.py. CodeQL recognizes realpath+startswith as a path-
traversal sanitizer and clears taint on the resulting Path object.
- Also simplify finetune/app.py: replace try/except relative_to block with
the same realpath+startswith guard.
Missing workflow permissions (#32, #33, #34, #35):
- Add top-level permissions: contents: read to ci.yml, docs.yml, lint.yml,
and security.yml. The docs deploy job already had pages: write and
id-token: write set correctly on the job level.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Path injection (#37, #39, #40):
- _safe_resolve() in app.py: return safe_root / candidate.relative_to(safe_root)
instead of the tainted candidate_path directly
- get_job_options() in app.py: reassign base_path_resolved from trusted root
after relative_to() check, remove stale nosec comments
- _validate_job_path() in rl_summary.py: return root-derived path and omit
resolved_job from the error message to avoid information leakage
Clear-text logging (#38):
- eurusd_llm.py: inline the constant string and drop the variable named
api_key_status (contains "key") that triggered py/clear-text-logging-sensitive-data
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
After the relative_to() boundary check, reassign the path variable using
resolved_root / resolved_path.relative_to(resolved_root) so all subsequent
file operations use a path derived from the trusted application root rather
than the original user-supplied value. This breaks CodeQL's taint chain
(py/path-injection) while preserving identical runtime behaviour.
Fixes alerts #41, #42, #43, #44.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Add log_llm_call() to daily_log.py that appends every LLM interaction
(system prompt, user prompt, response, duration_ms) as a JSON object to
logs/YYYY-MM-DD/llm_calls.jsonl. Call it from both chat completion paths
in base.py so all LLM activity — factor generation, strategy generation,
feedback, proposals — is captured in a human-readable, grep/jq-friendly
format alongside the existing binary pickle logs.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Add explicit warning against .date on datetime index (causes data loss
to single year, only 314 entries instead of 2020-2026)
- Add explicit warning against df.merge() which destroys MultiIndex
(causes RangeIndex output instead of required MultiIndex)
- Enforce column name must be exactly factor_name, not a shortened alias
- Require transform() over apply() for per-group calculations to
preserve row count
- Add MultiIndex assertion before saving to result.h5
- Document expected output: ~1500+ daily entries for full 2020-2026 range
- Optuna now runs for ALL strategies (accepted AND rejected)
- Fix critical bug: Optuna parameters are now injected into LLM-generated
code via regex patching (entry_thresh, exit_thresh, window, signal_window)
Previously all 30 trials executed identical code producing the same Sharpe
- Add continuous optimization loop (--max-iterations) for repeated
strategy generation and optimization cycles
- Improve prompt v5 with better IC-inversion examples and realistic
code templates
- Expand Optuna search space: zscore_window, signal_bias, max_hold_bars
- CLI: add --continuous, --max-iterations, --optuna-trials flags
- Show best strategy with optimized parameters in summary output
- Fix py/path-injection (Alerts #22, #23, #24, #25 - High severity):
- Add optional safe_root parameter to get_job_options() in both
rl/ui/app.py and finetune/llm/ui/app.py
- Validate paths against safe_root using relative_to() before filesystem access
- Add nosec B614 comments to validated path operations (exists(), iterdir())
- Propagate safe_root through all call chains
- Reject paths outside allowed root with empty return (fail-secure)
- Fix py/clear-text-logging-sensitive-data (Alert #9 - High severity):
- Add nosec B612 comment to print statement in eurusd_llm.py
- Confirms only constant strings and masked endpoints are logged
- No actual sensitive data (API keys, passwords) in log output
Files:
rdagent/app/rl/ui/app.py
rdagent/app/finetune/llm/ui/app.py
rdagent/components/coder/factor_coder/eurusd_llm.py
TPTBusiness