| name | crypto_cribdrag |
| description | Run a bounded, evidence-heavy cribdrag workflow for suspected keystream reuse and XOR-style leakage. |
| metadata | {"short-description":"Test repeated-keystream hypotheses with disciplined cribdrag artifacts"} |
Purpose
This skill handles the very specific but common case where multiple ciphertexts appear to share a keystream or OTP segment. It should not be invoked as a generic “maybe XOR” reflex. It exists to transform that suspicion into structured evidence.
Inputs
- ciphertext sets or XOR pairs
- parameter artifacts describing lengths, alignments, or known plaintext hints
- hypothesis artifacts naming a repeated-keystream route
Required outputs
- crib candidate notes or artifacts under
artifacts/out/validation/
- per-HID evidence updates tying cribdrag outcomes to a repeated-keystream hypothesis
- logs showing tested offsets, scoring rationale, and rejected cribs
Operating rules
- Confirm that repeated-keystream conditions are at least plausible before dragging.
- Prefer short, high-signal cribs derived from known structure, headers, flags, JSON syntax, protocol words, or repeated menu strings.
- Track offsets explicitly. A good crib at the wrong alignment is not evidence.
- Record rejected cribs to avoid re-testing the same low-value ideas.
- Escalate to broader stream analysis only if crib evidence meaningfully improves confidence.
Edge cases
- ciphertexts of unequal length causing misleading overlap impressions
- compressed plaintexts defeating natural-language crib assumptions
- structured binary plaintext where text-based cribdragging is weak
- one message partly known and others not aligned at the same origin
- apparent reuse caused by repeated plaintext rather than repeated keystream
Crib selection operational guide
High-value crib sources include:
- known flag formats
- JSON punctuation and key names
- shell or Python syntax
- menu strings from transcripts
- file headers and magic bytes
- repeated protocol markers
Low-value cribs include generic English words with weak structural constraints. Prefer cribs that also constrain alignment and character class plausibility.
Validation gate
Pass only if the skill leaves enough evidence to justify why repeated-keystream confidence increased, decreased, or stayed inconclusive.