How do GTFOBins' JSON API and its ATT&CK mapping file fit into an audit, and what does the technique mapping not tell you?
answer
- api.json for scripted lookups
- mitre.json is a Navigator layer
- mapped per function, not per host
- pin the snapshot you used
basics
~20 sapi.json holds the catalogue's functions, contexts and executables, so an audit can script the lookup. mitre.json is an ATT&CK Navigator layer built from each function's technique IDs; the mapping is per function and says nothing about your host.
solid answer
~40 sapi.json carries the function and context definitions and, for every executable, its functions and any alias, so a script can join a sudo and SUID inventory to the catalogue instead of searching by hand. mitre.json is an ATT&CK Navigator layer generated from the technique IDs attached to each function, for example T1059 for Shell and Command, T1105 for Download and T1548 for Privilege escalation. That mapping is coarse and community-assigned: it names a class of behaviour a function belongs to, not that anything happened on a host or that a technique is exposed there. Use it for reporting vocabulary and coverage conversations. Record the catalogue's commit or date, because both files change as entries are added.
go deeper
Know that the project publishes a JSON API and an ATT&CK Navigator layer.
Describe what api.json and mitre.json contain and that the technique IDs hang off functions.
Use them to script a repeatable fleet audit, pin the snapshot, and explain why the mapping is coarse and says nothing about a given host.
Decide how catalogue data enters the organisation's coverage reporting without letting entry counts stand in for exposure.
## What the two files hold The GTFOBins repository, read at commit `acd5246` on 2026-10-04, publishes two machine-readable files that the website links from its index page. - **`api.json`** (the "JSON API") contains the function definitions, the context definitions and an `executables` object. Each executable carries its `functions`, and where present a `comment` and an `alias`. It is built from the same data as the website. - **`mitre.json`** is a **MITRE ATT&CK Navigator layer**. It lists techniques and links each one back to the catalogue entries that map to it. The layer in the repository declares ATT&CK version 18, Navigator 5.2.0, layer format 4.5, the enterprise domain, and platforms Linux and macOS. ## The function-to-technique mapping The technique IDs are attached to **functions** in the project's data. Read on 2026-10-04: | Function | Technique IDs | |---|---| | Shell | T1059 | | Command | T1059 | | Reverse shell | T1059, T1071 | | Bind shell | T1059, T1071 | | File write | T1565 | | File read | T1005 | | Upload | T1041 | | Download | T1105 | | Library load | T1574 | | Privilege escalation | T1548 | | Inherit | none attached in the data file | What each ID names belongs to the ATT&CK framework itself; the point here is only that the catalogue's mapping is per function. ## How an audit uses them 1. **Script the join.** Load `api.json`, take the executable names from your sudo and set-user-ID inventories, and report the functions listed per executable. This is repeatable across a fleet, unlike a manual search. 2. **Report in shared vocabulary.** Findings written as "File read under Sudo, T1005" let a security team that tracks coverage against ATT&CK sort the audit alongside detection work. 3. **Feed detection planning.** The function an entry lists, and the technique IDs behind it, indicate which privileged executions deserve a baseline and which detections to test, without telling you how to write them. 4. **Pin the source.** Store the commit or download date with the report, since the entries and the mapping change. ## What the mapping does not tell you - **It is by function, not by host.** A technique ID on File read says what class of behaviour the function belongs to. It does not say the technique occurred, is possible on your host or is exposed through any grant you have. - **It is coarse and community-assigned.** Several functions share an ID, so the ID cannot distinguish them, and a different reader might map them differently. - **It does not count your risk.** A Navigator layer coloured by catalogue entries shows catalogue coverage, not how many of your accounts hold relevant grants. - **It ages.** The layer states ATT&CK version 18 in the clone; the framework moves on, and the file lags until its maintainers update it. - **Inherit carries no ID.** Treat it as a pointer to another entry. ## A scripted comparison, in outline A small script is enough. It reads the catalogue snapshot, reads a plain list of program names produced from the host's sudo and set-user-ID inventories, and prints one row per match: program, grant type, the functions listed under the matching context, and the snapshot date. Unmatched names are printed too, in a separate section headed "not in the catalogue, review by function", so nobody reads silence as clearance. Two cautions apply. Match on the program's name only after confirming the path, because a local script can share a name with a catalogued program. And keep the output with the snapshot it came from, so a later audit can tell a new grant from a new catalogue entry. ## Pitfalls Using the API as a verdict engine (matched, therefore vulnerable) repeats the misreading that the catalogue's own page rules out. Using the heatmap as a risk metric inflates coverage numbers because one catalogued program can appear under several functions. ## Summary The API makes the lookup repeatable and the ATT&CK file gives you shared vocabulary; neither turns a catalogue match into a measurement of exposure.
- Why is a Navigator heatmap built from the catalogue a poor measure of your own exposure?It counts catalogue entries per technique, not your grants. One program can appear under several functions and techniques, and none of it reflects which accounts, hosts or images actually hold a sudo rule, a set-user-ID bit or a capability. Exposure comes from joining your inventory to the entries.
- How do you keep a scripted lookup from drifting out of date?Pin the commit or download date of api.json in the script and in each report, refresh it on a schedule, and diff the executables and functions between versions so new entries that touch your grants are reviewed. Treat a missing newer entry as an open question, not as safety.
- Treats a technique ID as proof of exposure
- Uses the API match as an automatic vulnerability verdict
- Reads the Navigator heatmap as the host's risk score
- Assumes the mapping is complete and vendor-verified
- Never records which catalogue snapshot the audit used