Compute an HMAC signature for any message and secret using SHA-1, SHA-256, SHA-384 or SHA-512, with hex, base64 or base64url output.
Message authentication codes, locally. HMAC combines a hash function with a secret key so a recipient can verify both integrity and authenticity. Computed in your browser via the Web Crypto API.
An HMAC does not hide anything. It answers a different question: was this exact message produced by someone holding the shared secret, and has it been altered since? The output is a fixed-size tag you send alongside the message; the receiver recomputes it with the same key and compares. Matching tags mean the message is authentic and intact. Mismatched tags mean it is neither, and there is no way to tell which.
This makes HMAC the workhorse behind webhook signatures, API request signing, signed cookies and session tokens. Stripe, GitHub and most webhook providers sign their payloads this way so you can verify a request really came from them rather than from someone who guessed your endpoint URL.
The obvious construction — SHA256(key + message) — is broken against Merkle-Damgard hashes like SHA-1 and SHA-256. Those hashes process data in blocks and carry state forward, so an attacker who knows a valid digest can append data and compute a valid digest for the extended message without ever learning the key. This is the length-extension attack, and it is not theoretical; it has broken real APIs.
HMAC's nested construction, H(key XOR opad || H(key XOR ipad || message)), closes that hole. The two passes with different padded keys mean the final digest is not the raw internal state of a hash over attacker-influenced data. This is why you should always reach for HMAC rather than inventing your own keyed hash.
Choose SHA-256 unless something external forces your hand. SHA-1 is still safe inside HMAC because the construction does not rely on collision resistance, but new systems have no reason to use it and auditors will flag it. SHA-512 is not meaningfully stronger for this purpose and is sometimes faster on 64-bit hardware.
Two implementation details matter more than the hash choice. First, compare tags in constant time — a normal string comparison returns early on the first differing byte, which leaks enough timing information to forge a tag one byte at a time. Second, keys should be random bytes of at least the digest length, not a memorable passphrase; HMAC keys are not password hashes and get no stretching.