crypto-oracle-attacks
Skill26 starsChanged 32 days agoMITView on GitHub
Break weak symmetric crypto in real apps — decrypt or forge encrypted tokens/cookies without the key via padding oracles (CBC), ECB block shuffling, and hash-length-extension of homemade MACs. Load when the app hands you ciphertext you can tamper with and resend: an encrypted cookie/token, a "state"/"data" blob, an IV+ciphertext, a `H(secret‖msg)` signature, base64 that changes on edit.
2 other projects use a version of this.
Adopted by
Most recently taken up by
- NoorQureshi/SploitAgentword for word, first seen Oct 11, 2026
- NoorQureshi/security-skillsword for word, first seen Oct 11, 2026
Other versions2 projects use a version of this.
Word for word · 2
- NoorQureshi/security-skills★ 27100% same
- NoorQureshi/SploitAgent★ 26100% same
Edited · 0
None found.
Wrote their own version · 0
None found.
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository207 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- ad-adcsSkill
- ad-dacl-abuseSkill
- ad-delegation-abuseSkill
- ad-kerberoastingSkill
- ad-pivot-arsenalSkill
- ai-agent-tool-abuseSkill
- ai-insecure-output-handlingSkill
- ai-jailbreakSkill
- ai-llm-dosSkill
- ai-mcp-securitySkill
- ai-model-extractionSkill
- ai-prompt-injectionSkill
- ai-rag-poisoningSkill
- ai-sensitive-data-disclosureSkill
- ai-supply-chainSkill
- ai-system-prompt-leakageSkill
- api-auth-attacksSkill
- api-bolaSkill
- api-fuzzingSkill
- api-graphqlSkill
- api-grpcSkill
- api-mass-assignmentSkill
- api-mongo-agg-facet-bypassSkill
- api-testing-checklistSkill
- api-versioningSkill
- automation-browserSkill
- automation-nuclei-templatesSkill
- automation-recon-pipelineSkill
- cloud-azureSkill
- cloud-container-escapeSkill
- cloud-docker-api-abuseSkill
- cloud-docker-registrySkill
- cloud-gcpSkill
- cloud-iam-privescSkill
- cloud-imds-ssrfSkill
- cloud-kubernetesSkill
- cloud-s3-exposureSkill
- code-review-cicdSkill
- code-review-cppSkill
- code-review-dangerous-sinksSkill
- code-review-dotnetSkill
- code-review-goSkill
- code-review-iacSkill
- code-review-javaSkill
- code-review-methodologySkill
- code-review-nodejsSkill
- code-review-phpSkill
- code-review-pythonSkill
- code-review-rubySkill
- code-review-rustSkill
- code-review-secrets-detectionSkill
- code-review-soliditySkill
- crypto-rsa-attacksSkill
- ctf-cryptoSkill
- ctf-forensicsSkill
- ctf-methodologySkill
- ctf-pwnSkill
- ctf-revSkill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 43
Byte-at-a-time decryption: if your input is prepended to a secret and ECB-encrypted, recover the secret one byte per aligned block.
≈ “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.
Files
Oracle & symmetric-crypto attacks
When it applies
An app protects data with symmetric crypto you can poke — an encrypted session cookie, a "remember me"/state token, a signed-with-a-hash parameter — and it lets you submit modified ciphertext and observe a difference (an error vs. success, a padding error, different output). That observable difference is an oracle you can turn into full decrypt/forge, no key needed.
Why it works
Encryption without authentication is malleable: CBC leaks validity through padding errors, ECB
reveals structure and lets you rearrange blocks, and MAC = H(secret‖message) (Merkle–Damgård) can
be extended without knowing the secret. The key never leaks — the mode/construction does the work.
Method
- Fingerprint the token. Decode (base64/hex); measure length. A multiple of 8/16 bytes → a block cipher (CBC/ECB). Flip one byte and resend: a distinct "padding"/"decryption" error = CBC padding oracle; identical ciphertext blocks for identical plaintext = ECB.
- CBC padding oracle — when the server reveals valid vs. invalid padding: recover plaintext byte
by byte, and encrypt arbitrary plaintext (forge a token) by controlling the preceding block.
padbuster <url> <ciphertext> <blocksize> -cookies ..., or a Python oracle loop. This is the classic "decrypt/forge the admin cookie" attack. - ECB weaknesses — identical plaintext blocks → identical ciphertext:
- Cut-and-paste: rearrange whole blocks to splice a privileged value (e.g. move an aligned
role=adminblock into place). - Byte-at-a-time decryption: if your input is prepended to a secret and ECB-encrypted, recover the secret one byte per aligned block.
- Cut-and-paste: rearrange whole blocks to splice a privileged value (e.g. move an aligned
- Hash length extension — a signature of the form
H(secret‖data)(MD5/SHA1/SHA2) with knowndataand secret length: append your own data and compute a valid new MAC with hashpump (hashpump -s <sig> -d <data> -a <append> -k <keylen>), bypassing the "integrity" check. - Weak/predictable tokens — if it turns out not to be real crypto: ECB pattern, a static IV, XOR with a fixed keystream, or a guessable PRNG seed (timestamp) → derive/forge directly.
- Forge and use — mint the token that says what you want (admin, another user) and replay it;
validate the access gained with
reporting-triage-validation.
Gotchas
- Authenticated encryption (AES-GCM, encrypt-then-HMAC) kills these — a padding oracle needs an unauthenticated CBC error signal; if tampering just fails uniformly, move on.
- The oracle can be subtle — a timing difference, a 500 vs 403, or a different redirect counts; diff responses carefully before concluding "not vulnerable".
- Block alignment is everything for ECB cut-and-paste and byte-at-a-time — pad your input to land targets on block boundaries.
- HMAC (not
H(secret‖msg)) is not extendable — length extension only hits raw Merkle–Damgård concatenation MACs.
Verify success
A token you decrypted or forged that the application accepts as another/privileged user (or recovered plaintext of a value you couldn't read), reproduced from a clean session.
References
Vaudenay padding-oracle; padbuster; the Cryptopals set (mechanisms); hashpump.
Related: web-auth-jwt, crypto-rsa-attacks, web-idor.
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
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.

