SATAgent

Run the pure-LLM variant of the QLCoder workflow for both CVE samples and local Web App repositories. Supports multi-profile taint-flow analysis with source/...

Security Scan

The skill appears to implement a QLCoder 'pure-LLM' taint/triage workflow and mostly matches that purpose, but there are inconsistencies (mismatched skill naming and hard-coded developer paths) and the bundled script runs local filesystem scans and subprocesses — review before trusting it on sensitive code.

Purpose & Capability

The SKILL.md and bundled files implement a QLCoder pure-LLM taint-flow analysis (Java/Python web) which matches the described capability, but the registry metadata/name ('SATAgent') does not match the SKILL.md name/slug (qlcoder-pure-llm / coder-pure-llm). This naming mismatch and leftover absolute paths (e.g., /Users/aibot/...) are incongruent and suggest sloppy packaging or repackaging.

Instruction Scope

Instructions intentionally read repository files, build manifests, and run local analysis; the bundled script performs wide file-system scanning (rglob over repo), pattern matching, writes JSON artifacts, and calls subprocess.run to invoke external commands (e.g., python -m qlcoder.cli). Those actions are consistent with code-analysis purpose, but SKILL.md contains hard-coded absolute developer paths and encourages running the wrapper which will search the entire repo — a significant scope that should be reviewed by the user.

Install Mechanism

No install spec or remote downloads are present; the skill is instruction-only with a bundled Python script. Nothing external is fetched during install by the skill package itself.

Credentials

The skill declares no required environment variables or credentials and the code does not require external tokens. It reads local files and may invoke local Python modules; no unnecessary secrets access is requested.

Persistence & Privilege

always:false and no special persistence or modifications to other skills are requested. The skill can be invoked by the model (normal default). The primary risk is the skill's ability to read workspace/repo files and run subprocesses when executed.

What to consider before installing

This skill runs local repository-wide scanning and executes subprocesses (it invokes python -m qlcoder.cli and similar). Before installing or running: 1) Confirm the QLCoder project you expect is present (qlcoder/cli.py) and that you trust that code; 2) Inspect scripts/run_pure_llm.py (already bundled) to ensure there are no hidden network calls or unexpected commands beyond the visible filesystem scanning and subprocess usage; 3) Note the SKILL.md contains developer-specific absolute paths (/Users/aibot/...), update those to your environment or run from a copy; 4) Run the skill in an isolated sandbox or on a non-sensitive copy of the repository first; 5) If you need stronger assurance, ask the publisher to fix the naming/packaging inconsistencies and remove hard-coded paths. These discrepancies (name mismatch and hard-coded paths) are the main reasons this analysis is flagged as suspicious rather than benign.

QLCoder Pure LLM

Use this skill to drive pure-llm mode in the local QLCoder project with two workflows:

• CVE sample analysis from manifest (patch/advisory-driven).
• Local repository taint-flow typing for Web App codebases (Java Web, Python Web).

The local workflow now also emits a triaged security finding summary (taint_findings.analysis.*) focused on XSS / SQLi / broken-access-control classes.

Quick Start

1. Confirm the workspace contains qlcoder/cli.py.
2. Prefer the helper script: python /Users/aibot/.codex/skills/coder-pure-llm/scripts/run_pure_llm.py --workspace /path/to/workspace
3. For CVE mode, optionally pass --project-slug <slug> or --batch-limit N.
4. For local Web App mode, pass: --repo-path /path/to/repo --app-profile auto
5. Read generated artifacts under runs/.../pure-llm/.

Workflow

A. CVE Manifest Mode

1. Reuse an existing manifest under datasets/manifests/ when possible.
2. If missing, build base manifest: python -m qlcoder.cli build-manifest --base-only --allow-manual-override --manual-base-ref master
3. Run single sample (run-one-pure-llm) or batch (run-manifest-pure-llm).
4. Summarize:
   • summary
   • source_hypotheses
   • sink_hypotheses
   • sanitizer_hypotheses
   • draft_query_strategy
   • draft_query_text

B. Local Web App Mode (Java/Python)

1. Run: python /Users/aibot/.codex/skills/coder-pure-llm/scripts/run_pure_llm.py --workspace /path/to/workspace --repo-path /path/to/repo --app-profile auto
2. --app-profile values:
   • auto: detect by repository file distribution; mixed repositories can run both Java and Python profiles.
   • java-web: force Java Web source/sink/sanitizer typing.
   • python-web: force Python Web source/sink/sanitizer typing.
3. First review taint_findings.analysis.json/md for prioritized findings and severity.
4. Then use taint_profile.analysis.json/md for deeper source/sink/sanitizer trace evidence.
5. Treat these results as taint-flow leads for manual trace confirmation.

Source/Sink Type Reference

See references/taint_profiles.md for the taxonomy used by the local Web App profile report.

Output Contract

CVE manifest mode artifacts

• runs/<project_slug>/pure-llm/iteration_00.prompt.md
• runs/<project_slug>/pure-llm/iteration_00.analysis.json
• runs/<project_slug>/pure-llm/iteration_00.analysis.md
• runs/<project_slug>/pure-llm/iteration_01.analysis.json
• runs/<project_slug>/pure-llm-summary.json

Local Web App mode artifacts

• runs/local-<analysis_name_or_repo>/pure-llm/analysis.json
• runs/local-<analysis_name_or_repo>/pure-llm/analysis.md
• runs/local-<analysis_name_or_repo>/pure-llm/taint_profile.analysis.json
• runs/local-<analysis_name_or_repo>/pure-llm/taint_profile.analysis.md
• runs/local-<analysis_name_or_repo>/pure-llm/taint_findings.analysis.json
• runs/local-<analysis_name_or_repo>/pure-llm/taint_findings.analysis.md

Treat JSON artifacts as source of truth. Use Markdown for human-readable review.

Notes

• Do not claim CodeQL validation in this mode.
• Missing patch or advisory files are tolerated; the workflow should still proceed using whatever metadata is available.
• Taint profile output is pattern-based candidate analysis, not full dataflow proof.
• Java ${...} sink typing is de-noised to mapper SQL context to avoid common pom.xml/config placeholder false positives.
• The findings summary is heuristic triage; always verify exploitability with targeted trace/tests.
• For detailed command patterns and wrapper usage, read references/workflow.md.