agent-passport-system

Repository on GitHub, which provides public git repositories with URLs.

Repository on GitHub, which uses git. git can track the changes, who made them, and when they were made.

879 commits across 10 tagged releases. Ordinary incremental commits exist between every release tag, not squashed snapshots limited to final versions.

Repository on GitHub, which uses git. git is distributed.

package.json version is bumped for every release, and npm's registry rejects republishing an already-used version number, so every published version is guaranteed unique.

Releases are tagged in git following the vMAJOR.MINOR.PATCH convention (v1.5.1 through v3.1.0 confirmed in the tag history).

Non-trivial release notes file in repository: https://github.com/aeoess/agent-passport-system/blob/main/CHANGELOG.md.

No release has fixed a publicly known, CVE-assigned vulnerability to date.

Non-trivial SECURITY[.md] file found file in repository: https://github.com/aeoess/agent-passport-system/blob/main/SECURITY.md. [osps_do_02_01]

GitHub Issues is used actively for tracking individual issues. https://github.com/aeoess/agent-passport-system/issues

No issues in the 2-12 month qualifying window are labeled as bug reports; activity in that window is enhancement, RFC, and design-discussion proposals, the large majority of which have maintainer replies.

6 of 9 enhancement-labeled issues opened in the 2-12 month window have maintainer replies (67%), exceeding the 50% threshold.

GitHub Issues is a public, URL-addressable, searchable archive of reports and responses. https://github.com/aeoess/agent-passport-system/issues

SECURITY.md publishes the process: report to security@aeoess.com, with a stated 48-hour acknowledgment target and 7-day fix-timeline target. https://github.com/aeoess/agent-passport-system/blob/main/SECURITY.md

SECURITY.md specifies a private channel (security@aeoess.com) and explicitly instructs reporters not to open a public issue for security vulnerabilities. https://github.com/aeoess/agent-passport-system/blob/main/SECURITY.md

No vulnerability reports have been received in the last 6 months

npm run build (tsc && chmod +x dist/src/cli/index.js) automatically rebuilds the software from source in one command.

Uses the TypeScript compiler (tsc), the standard build tool for the language.

tsc (TypeScript) is itself FLOSS; no proprietary tooling is required to build.

npm test runs an extensive automated FLOSS test suite (tsx --test across hundreds of test files), documented in package.json and CONTRIBUTING.md.
https://github.com/aeoess/agent-passport-system/blob/main/CONTRIBUTING.md

npm test invokes the full suite, the standard convention for Node projects; cargo test covers the Rust workspace.
https://github.com/aeoess/agent-passport-system/blob/main/package.json

Four GitHub Actions workflows run automated tests on changes: aps-conformance.yml, check-drift.yml, contributor-check.yml, cross-impl-jcs.yml. https://github.com/aeoess/agent-passport-system/tree/main/.github/workflows

CONTRIBUTING.md's contribution checklist states explicitly, for feature additions: "Tests cover the new behavior at the same density as adjacent code." https://github.com/aeoess/agent-passport-system/blob/main/CONTRIBUTING.md

Commit 42b1fd7 (2026-07-04) added a major new feature (word digest handles and read-fidelity receipts) and landed 544 and 375 lines of new tests in the same commit as the new source, following the stated policy.

The policy is written directly into CONTRIBUTING.md's contribution checklist, not just followed informally.

TypeScript strict mode is enabled (strict: true in tsconfig.json).

CONTRIBUTING.md states tsc --noEmit (strict mode, no new any in public surface) must be clean for a PR to be mergeable.

TypeScript strict mode is enabled project-wide. Additional opt-in strictness (e.g. noUncheckedIndexedAccess) and a supplementary linter are not yet in place.

SECURITY.md documents an explicit threat model. Beyond that, today's work is direct evidence: identified and correctly remediated five distinct real vulnerability classes (prototype pollution, SSRF, command injection, ReDoS, log injection) across the codebase, with empirical verification for each, not just pattern-matching.

Beyond SECURITY.md's named threat classes, demonstrated directly: found and fixed prototype pollution (Object.create(null) plus key rejection), SSRF (BlockList-based guard with DNS-rebinding protection), shell command injection (execFileSync over string interpolation), catastrophic regex backtracking (empirically timed, 16.7s to 1ms), and log injection (JSON.stringify escaping). Each fix was verified, not assumed.
https://github.com/aeoess/agent-passport-system/security/code-scanning

Ed25519 (RFC 8032), via the audited @noble/ed25519 library, is the core identity and signing primitive.

Cryptography is delegated to @noble/ed25519 and libsodium-wrappers, both dedicated crypto libraries; no custom cryptographic primitives are implemented in-house.

@noble/ed25519 and libsodium-wrappers are both FLOSS (MIT and ISC respectively).

Ed25519 uses a fixed 256-bit key with no smaller-keylength configuration option, exceeding NIST's 2030 minimums by construction; there is no weaker-keylength mode to need to disable.

The default identity and signing mechanism is Ed25519, not any of the listed broken algorithms.

Ed25519 uses SHA-512 internally; no default mechanism depends on SHA-1 or similarly weakened algorithms.

APS implements signing and identity, not a key-agreement/session protocol. Transport-level forward secrecy is provided by standard TLS, outside code this project implements.

APS does not store user passwords. Identity is Ed25519 key-based (did:key, did:web, SPIFFE, OAuth); there is no password-authentication surface.

Key generation goes through @noble/ed25519 and libsodium-wrappers, both of which use a cryptographically secure RNG by default.

Distribution channels use HTTPS exclusively. [osps_br_03_02]

All distribution (npm, GitHub) is HTTPS-only; no hash is fetched over plain http.

No publicly known vulnerability has been reported against the project, so there is nothing unpatched.

No critical vulnerability has been reported to date; nothing has required a response-time track record yet.

Checked directly: no .env, .pem, .key, or private-key files are tracked in git, no PEM key blocks or secret-shaped strings exist in tracked source.GitHub's native secret scanning and push protection are both enabled on the repository, actively blocking any push containing a known secret pattern.

CodeQL (security-extended) has been run against the codebase. https://github.com/aeoess/agent-passport-system/actions/runs/28966837754

security-extended specifically targets vulnerability classes (injection, security-check bypass, ReDoS), not just style rules.

All 14 open alerts resolved: 5 fixed with verified code changes (prototype pollution, command injection, SSRF, one test tightened), 9 dismissed as confirmed false positives, each with traced evidence. Combined with the 6 fixed and 4 dismissed earlier this session, 0 alerts remain open, confirmed by re-running the tool after every change.
https://github.com/aeoess/agent-passport-system/security/code-scanning

Coverage-guided fuzzing (seven Jazzer.js harnesses) runs via ClusterFuzzLite on every pull request and on a scheduled batch; releases are tagged from main that has passed these runs. The campaign has found and fixed two real bugs, kept as corpus regression seeds. https://github.com/aeoess/agent-passport-system/blob/main/fuzz/FINDINGS.md

The project's compiled component (aps-verifier-core) is written in Rust, a memory-safe language, not C or C++. No memory-unsafe language is used.

The test suite runs via Node's native test runner with extensive runtime assertions checking expected behavior across the protocol's layers.

No dynamic analysis tool is in use, so nothing has been discovered via this mechanism to require fixing.