GitHub Security Lab Taskflows
This repository contains example taskflows to use with the SecLab Taskflow Agent, as well as the custom MCP servers that are needed to run the taskflows.
Quick start
- Go to https://github.com/GitHubSecurityLab/seclab-taskflows and start a codespace.
- Wait a few minutes for the codespace to start. It's ready when you see
(.venv)before the prompt in the terminal. - Run the demo:
python -m seclab_taskflow_agent -t seclab_taskflows.taskflows.audit.ghsa_variant_analysis_demo -g repo=github/cmark-gfm -g ghsa=GHSA-c944-cv5f-hpvr
Now try running our auditing taskflows on one of your projects. Here, we're using the OWASP Juice Shop as an example:
./scripts/audit/run_audit.sh juice-shop/juice-shop
⚠️ Note: the auditing taskflows can take several hours to run, especially on larger projects, and make a lot of AI requests, which can cost a non-trivial amount of money. By default, the taskflows use the Copilot API. You can find out more information about Copilot Billing here. Alternatively, you could use a different AI API by setting the
AI_API_ENDPOINTenvironment variable, as explained here.
The results of the audit are written to an SQLite database, which is opened automatically in an SQLite viewer at the end of the run. The results are in the table named "audit_result". The table has a column named "has_vulnerability", with checkmarks in the rows that are most likely to be genuine vulnerabilities.
Mobile application support
The audit workflow recognizes Android, iOS, and cross-platform mobile applications alongside web applications and libraries. Mobile entry-point collection records exported Android components, deep links, URL schemes, Universal Links, app extensions, and WebView JavaScript bridges, including their permissions, export status, and input filters.
The audit runner collects this metadata before classification, which applies mobile-specific
guidance to components marked as mobile applications. The existing --advisory and -m options also apply to mobile runs.
Both mobile entry-point taskflows use the container-backed source-access toolbox and require Docker.
Mobile metadata is stored in the mobile_entry_point table. Existing context databases automatically gain the nullable
application.is_mobile_app column without deleting existing results.
Running with docker script
We recommend running taskflows in a sandboxed environment. GitHub Codespaces are convenient, or if you prefer you can use the script run_seclab_agent.sh to run a docker container of the seclab-taskflow-agent as outlined here. Note that this script needs to be run from the main directory of the repo, and the .env file with the environment variables for the custom MCP servers to store data needs to be in the same directory.
First, create a .env file in the main directory of the repo. For run_seclab_agent.sh you can use:
MEMCACHE_STATE_DIR=/app/data
CODEQL_DBS_BASE_PATH=/app/data
DATA_DIR=/app/data
LOG_DIR=/app/logs
The MEMCACHE_STATE_DIR is needed to persist some intermediate data in the memcache, DATA_DIR is needed for various mcp servers to store intermediate results, and LOG_DIR is used to store log files generated by the servers. These can be set in a .env file in the main directory. If no environment variables are set for the custom MCP servers, relevant folders will be created automatically. The location depends on the platform, and is set by platformdirs.
In addition, AI API endpoints and secrets also need to be configured via environment variables or Codespace secrets. In particular, the environment variables AI_API_TOKEN and AI_API_ENDPOINT need to be set to the appropriate AI API endpoints and credentials. If not set, the default AI_API_ENDPOINT is GitHub Copilot:
AI_API_ENDPOINT="https://api.githubcopilot.com"
In addition, the GH_TOKEN environment variable also needs to be set to allow interaction with the GitHub API, such as fetching content, creating issues etc.
Individual taskflows may need additional setup, please refer to the README.md in the relevant subdirectories for further requirements.
After setting the relevant env vars, run an example taskflow with:
./scripts/run_seclab_agent.sh -t seclab_taskflows.taskflows.audit.ghsa_variant_analysis_demo -g repo=github/cmark-gfm -g ghsa=GHSA-c944-cv5f-hpvr
Background
SecLab Taskflows is a companion repository to the SecLab Taskflow Agent repository. SecLab Taskflow Agent is an experimental agentic framework maintained by GitHub Security Lab. This repository provides example taskflows and supporting resources for use with the SecLab Taskflow Agent. We are using the agent and these taskflows to experiment with using AI Agents for security purposes, such as auditing code for vulnerabilities or triaging issues.
We'd love to hear your feedback. Please create an issue to send us a feature request or bug report. We also welcome pull requests (see our contribution guidelines for more information if you wish to contribute).
Requirements
- Python >= 3.11 for local execution
- Docker for container-backed taskflows
The container images from ghcr.io/githubsecuritylab/ use the latest tag.
Because this tag is mutable, a locally cached image may become outdated and
Docker may continue using it instead of the current registry image. Remove the
local images before running a taskflow when you need to ensure that Docker pulls
the latest versions:
docker rmi -f $(docker images --format '{{.Repository}}:{{.Tag}}' | grep '^ghcr.io/githubsecuritylab/')
License
This project is licensed under the terms of the MIT license. Please refer to the LICENSE file for the full terms.
Maintainers
Support
Metadata
Release files for seclab-taskflows 0.2.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| seclab_taskflows-0.2.0.tar.gz | 108.4 kB | With provenance | Details for seclab_taskflows-0.2.0.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| seclab_taskflows-0.2.0-py3-none-any.whl | Python 3 | none | any | Details for seclab_taskflows-0.2.0-py3-none-any.whl |
Total release size: 273.4 kB
Release files / seclab_taskflows-0.2.0.tar.gz
| Download URL | seclab_taskflows-0.2.0.tar.gz |
|---|---|
| Size | 108.4 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
51a615695ef63ce1cc565e53482ac68a592e8c113d7e55c2135d6a513ba6ef9e
|
|
BLAKE2b-256 checksum How to use checksums |
c18f75fdb09e6fda105f0b0e1b831951ea918f5c25531d6f450cf46f5700fb02
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Oct 21, 2026.
Transparency logRelease files / seclab_taskflows-0.2.0-py3-none-any.whl
| Download URL | seclab_taskflows-0.2.0-py3-none-any.whl |
|---|---|
| Size | 165.0 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
866508c772a0f32268317f9ff627941e55e8de3ab2b9c75ca17b32b86289aa5d
|
|
BLAKE2b-256 checksum How to use checksums |
4fd969294180161c88138ca708c5a6139ff905019d73d434020ffa870690075c
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Oct 21, 2026.
Transparency logSecurity for seclab-taskflows 0.2.0
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
Full provenance
Both files in this release have been uploaded with provenance metadata.
Attestation for seclab_taskflows-0.2.0.tar.gz
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Oct 21, 2026.
Transparency logA note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.