CF debugging
Runs a hypothesis-driven debugging loop across any language or binary, escalating to orthogonal oracle angles and locking the fix with a failing test. Use for crashes, silent failures, hangs, wrong responses, memory leaks, async misbehavior, or reverse engineering.
Runs a hypothesis-driven debugging loop across any language or binary, escalating to orthogonal oracle angles and locking the fix with a failing test.
As a process F 39/100 · Will not run — References files that are not bundled: references/runtimes/<runtime>.md, references/tools/*.md
What is at stake
Medium-severity findings: the skill is probably honest, but read what alarmed the scanner.
Below is the worst case for this category. The finding here is medium: the guard saw a sign, not a proof.
The skill contains commands that delete files, rewrite disks or execute code fetched from the network. The agent may run them without asking if it believes the instructions require it.
Replace destructive commands with safe equivalents that ask for confirmation, scope them to one folder, and stop piping curl into a shell: pin a version and a checksum.
How to improve
- The text references files that are not there: add them or drop the references.
- Your own cases (evals/evals.json, 4–6 real requests with expected answers): the full check would then run those instead of a model-drafted suite.
- A spec.yaml with trigger phrases and assertions — a behaviour contract for CI; `skilltest init` writes a template.
Guard findings · 7
✓ No critical or high findings
Medium and low: 7
-
medium Dangerous commands
cmd-privilegereferences/tools/pwntools.md:262Privilege escalation / world-writable permissions# echo 2 | sudo tee /proc/sys/kernel/randomize_va_space # Linux default
-
low Secrets in code
secret-high-entropy-tokenreferences/runtimes/bundled-js-binary.md:50High-entropy token-like string (may be an id, hash or a credential) (documentation table row)| `NODE_SEA_BLOB` + `NODE…2:1` | **Node SEA** (`node --build-sea` or `--experimental-sea-config`) | Plaintext JS, or V8 code cache (when `useCodeCache: true`),
table -
low Secrets in code
secret-high-entropy-tokenreferences/runtimes/bundled-js-binary.md:83High-entropy token-like string (may be an id, hash or a credential) (quoted — discussed, not commanded)- The fuse string `NODE…2:1` (with trailing `:1` indicating injected; `:0` means a copy of the node binary that has not yet had a blob injected)
quoted -
low Secrets in code
secret-high-entropy-tokenreferences/runtimes/bundled-js-binary.md:87High-entropy token-like string (may be an id, hash or a credential) (detector / deny-list definition)LC_ALL=C grep -aob 'NODE…2:1' ./target | head -1
detector -
low Dangerous commands
cmd-privilegereferences/runtimes/native-binary.md:280Privilege escalation / world-writable permissions (negated — the text forbids it)1. **SIP / restricted process** (target has the `__RESTRICT,__restrict` section, is setuid/setgid, or is a platform/Apple-signed binary): dyld unconditionally strips all `DYLD_*` env vars before the p
negated -
low Risky intent
intent-offensive-securityreferences/tools/ghidra.md:7Offensive-security / dual-use content (legitimate for authorised testing; review intended use) (detector / deny-list definition)**Use Ghidra when**: third-party closed-source libs, malware analysis, vendored binaries whose behavior contradicts docs, CTF challenges, firmware, any time you need to read compiled code.
detector -
low Dangerous commands
cmd-privilegereferences/tools/pwntools.md:245Privilege escalation / world-writable permissions (detector / deny-list definition)- **ASLR on local runs**: turn off for reproducibility during debugging: `echo 0 | sudo tee /proc/sys/kernel/randomize_va_space` (remember to revert — journal this!).
detector
Files scanned: 25. Evidence is masked. Grey chips explain why severity was lowered.
Against the Agent Skills spec
- warning
missing-refreference to a missing file: references/runtimes/<runtime>.md - warning
missing-refreference to a missing file: references/tools/*.md
Process rating: all ten parameters 39/100
- 0Tools and files. 2 referenced file(s) missing: references/runtimes/<runtime>.md, references/tools/*.md
- 0Result and completion. Does not say what the result is
- 0Inputs and preconditions. Does not say what the process needs to start
- 0Progress reporting. Says nothing while it works
- 20When it triggers. No condition that starts the skill
- 30Running it twice. 7 mutating operations with no state check
- 60Failures and branches. 2 branches
- 100Steps. 16 steps
- 100Consistency. Name and required fields are in place
- 100Execution cost. Instruction body is 3038 tokens
- medium Safety rules and hard prohibitions inside a skill: they belong in the system prompt, here they protect nothing
- low The response is described with custom markup (3 tags): a typed call is more reliable
Everything here is measured from the skill text rather than judged by a model, so the numbers are checkable. A parameter weighs more when it is a more common reason for the process to stall.
Quality signals
- +5Description has no quoted example phrases that should trigger the skill
- +4Description does not say when NOT to use the skill (false activations)
- +3Output format is not stated: the model decides each time
- +4No input/output examples
- -218 emoji in the instructions: noise for the model
- +1No license
- +2Single-language instructions
- +3Description length 265: enough signal without eating the budget
- +4Structure: 8 headings
- +3Step-by-step instructions: 16 items
Quality base 70; lint remarks subtract, signals add up to 100. Result: 69.