Skip to main content

decode-a-jwt-safely

Decode a JWT without confusing readable claims with verified identity or authorization. Use when inspecting token structure, metadata, timing, audience, or signature requirements.

Source facts

Repository
protosphinx/sphinxstack
Last source activity
July 29, 2026 at 13:55
Detected SKILL.md language
English
Stars
3
Forks
0

Install options

The review-first prompt is selected by default. You can switch to a direct command or download a local copy.

Review the source files

Read SKILL.md and any companion files shown by SkillsMP before deciding whether to install.

File Explorer
2 files

Showing SKILL.md

SKILL.md
Source instructions · Read-only preview
name
decode-a-jwt-safely
category
code
description
Decode a JWT without confusing readable claims with verified identity or authorization. Use when inspecting token structure, metadata, timing, audience, or signature requirements.
# decode-a-jwt-safely ## When to use - Use for bounded debugging with an authorized redacted token. - Never paste live tokens into third-party tools or treat decoding as verification. ## Preconditions - Record token source, environment, expected issuer, audience, algorithm, and current time. ## Procedure 1. Work offline on a redacted or expired fixture where possible. 2. Parse header and payload without executing embedded content. 3. Inspect algorithm, key ID, issuer, audience, subject, scopes, issue, not-before, expiry, nonce, and token ID. 4. Distinguish malformed, decoded, signature-verified, claim-validated, and authorized states. 5. Verify signature only with trusted issuer metadata and allowed algorithms. 6. Validate time, issuer, audience, nonce, replay, and application authorization separately. ## Failure plan - Reject ambiguous algorithm, key, issuer, audience, or clock state. ## Worked example A token decodes correctly but uses the wrong audience, so the service rejects it before authorization. ## Done - A JWT inspection report records safe metadata, verification state, claim checks, and authorization boundary - Redaction, issuer, algorithm, key, audience, time, nonce, replay, and fixture checks verify handling
View on GitHub