agentleFS
BrownFineSecurityBrownFineSecurity/iothackbot/skills/jtagprobe/SKILL.md

jtagprobe

Skill847 starsChanged 5 months agoMITView on GitHub

Probe IoT/embedded targets for exposed SWD/JTAG debug interfaces using a SEGGER J-Link. Detects whether debug is OPEN, LOCKED (readout-protected), or DEAD (fused off). Use when assessing whether a target's on-chip debug port can be reached, identifying the silicon vendor from DPIDR/IDCODE, and confirming halt+memory access for full debugger control.

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 3 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    If JLinkExe cannot be found at all, tell the user to install SEGGER J-Link software from segger.com.

    line 24
  2. Very specific:

    Default — sweep SWD then JTAG at 4000/1000/100 kHz, halt, read memory, classify:

    line 28
  3. Distinctive:

    Run a JTAG chain auto-scan as last resort

    line 50
  4. Very specific:

    If the user knows the chip, pass --device for a more accurate halt/memory test.

    line 60
  5. Distinctive:

    If the target is in a state where halting would crash an active firmware path you care about (rare in pentests, common in live systems), …

    line 74
  6. Distinctive:

    Default assumption: they want both SWD and JTAG checked, full halt+memory test, and evidence captured.

    line 123
  7. Very specific:

    If JLinkExe isn't on PATH, stop and report that the SEGGER tools aren't installed.

    line 128
  8. Very specific:

    Does not attempt unlock / mass-erase.

    line 135

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill 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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Jtagprobe — SWD/JTAG Debug Interface Tester

You are helping the user determine whether a target's on-chip debug interface is exposed via SWD or JTAG, using the jtagprobe tool. This drives a SEGGER J-Link physically wired to the target.

What the tool tests

Three nested access layers are checked, and the target is classified into one of:

  • OPEN — DP responds, CPU halts, memory reads return plausible data. Full debugger control. Critical finding.
  • LOCKED — DP/IDCODE accessible but memory reads fail or return readout-protection sentinels (0xFFFFFFFF). Indicates STM32 RDP, NXP CRP, Nordic APPROTECT, etc. are engaged. Still a finding — the port should not respond at all in production.
  • DEAD — No DP/IDCODE response on any tested interface/speed. Debug fused off, pins not wired, or wrong target.

Prerequisites

  • A SEGGER J-Link (any variant) connected via USB
  • JLinkExe on PATH — verify with which JLinkExe. If it is installed but not on PATH, point the tool at it with --jlink-binary /path/to/JLinkExe instead of relying on PATH.
  • Target wired to the J-Link 20-pin (or 10-pin Cortex Debug) header. Confirm SWDIO/SWCLK or TDI/TDO/TMS/TCK identification before energizing the target.

If JLinkExe cannot be found at all, tell the user to install SEGGER J-Link software from segger.com. Do not attempt to install it without explicit approval.

Basic usage

Default — sweep SWD then JTAG at 4000/1000/100 kHz, halt, read memory, classify:

jtagprobe

Save per-attempt JLinkExe logs as evidence (recommended for pentest writeups):

jtagprobe --evidence-dir ./evidence/jtagprobe-$(date +%Y%m%d-%H%M%S)

JSON for chaining:

jtagprobe --format json

Common workflows

1. Unknown target, unknown protocol

Just run with defaults. The tool will:

  1. Try SWD at 4 MHz → 1 MHz → 100 kHz
  2. Fall back to JTAG with the same speed sweep
  3. Run a JTAG chain auto-scan as last resort
  4. Identify vendor from DPIDR/IDCODE JEP106 designer field
  5. Halt CPU and read memory to confirm access level
jtagprobe --evidence-dir ./evidence

2. Known target — pass the device name

If the user knows the chip, pass --device for a more accurate halt/memory test. Use the same device strings J-Link accepts (STM32F407VG, nRF52840_xxAA, MK64FN1M0xxx12, etc.):

jtagprobe --device STM32F407VG

3. Slow targets / long traces / level-shifted boards

Some pirate-flagged boards or long ribbon cables need a slower clock. Limit the sweep:

jtagprobe --speeds 1000,100,10

4. Layer-1 only (no halt)

If the target is in a state where halting would crash an active firmware path you care about (rare in pentests, common in live systems), stop after the connect probe:

jtagprobe --skip-memory

5. SWD only or JTAG only

jtagprobe --interfaces SWD
jtagprobe --interfaces JTAG --speeds 4000,1000

Interpreting the output

The text format leads with the classification and reason:

CLASSIFICATION: LOCKED
DP/IDCODE accessible but CPU halt or memory read failed. Typical of RDP / CRP / APPROTECT engaged.

Vendor: STMicroelectronics
  SW-DP DPIDR=0x2BA01477 partno=0xBA version=2 designer_identity=0x20

Access test:
  Halted: True
  CPUID @ 0xE000ED00 = 0x410FC241
  0x08000000: 0xFFFFFFFF 0xFFFFFFFF 0xFFFFFFFF 0xFFFFFFFF [all-0xFF, possible RDP]

Protection hint: STM32 RDP Level 1/2 (see RM, FLASH_OPTR bits 15:8).

Key signals:

  • A non-zero DPIDR / IDCODE means the silicon answered. Even alone this is reportable.
  • 0xFFFFFFFF flash reads after a successful halt = readout protection. Capture the DPIDR and document the protection mechanism.
  • A "plausible vector table" (initial SP in SRAM range, reset vector with Thumb bit set) is the strongest signal of OPEN access — call this out in writeups.

Writeup-relevant detail

For a pentest finding under CWE-1191 (improper access control on debug interface) or CWE-1244 (asset exposed via debug):

  • Command run: full jtagprobe invocation
  • Classification + reason from the output
  • DPIDR / IDCODE raw value and decoded vendor
  • For OPEN: vector table words proving memory was read
  • For LOCKED: the all-0xFF read proving readout protection is the only line of defense (and that the port itself is still exposed)
  • Evidence files from --evidence-dir for the appendix

When the user says "test for JTAG"

Default assumption: they want both SWD and JTAG checked, full halt+memory test, and evidence captured. Run:

jtagprobe --evidence-dir ./evidence/jtagprobe-$(date +%Y%m%d-%H%M%S)

If JLinkExe isn't on PATH, stop and report that the SEGGER tools aren't installed.

Limitations

  • Requires physical access to the debug header and a J-Link probe wired up. This is not a network or pcap-based check.
  • Generic Cortex-M device profiles are used when no --device is passed. Halt/memory access may succeed under a generic device even when the vendor-specific erase/unlock would not.
  • JLinkExe stdout parsing is regex-based. If SEGGER changes the format in a future release the parser may need updating — --format json shows what was extracted.
  • Does not attempt unlock / mass-erase. That is destructive and out of scope for a probe. Use the vendor's bootrom or unlock commands separately with explicit authorization.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.