Every check runs real code against this repository and records what it observed.
| ID | Area | Check | Result | Observed |
|---|---|---|---|---|
| A01 | agents | Agent manifest, skills and Claude role files agree (20 roles) | PASS | 20 manifest entries, 20 SKILL.md files, 20 role files; frontmatter and 7 required sections present in each agents=20 |
| A02 | agents | Every skill forbids autonomous production action and defers to humans | PASS | all 20 skills contain the no-autonomy, human-owner and untrusted-input rules |
| A03 | contracts | Schemas parse; generated report and every event validate | PASS | report + 9 events + agent contract validate |
| A04 | determinism | Identical input produces an identical report ID and bytes | PASS | reportId inc_08ab6bbd890e21fe62bd |
| A05 | acceptance | Spec acceptance: time zones sort, correlation stays unconfirmed, absent metrics stay unknown, claims cite sources | PASS | {"sorted":true,"mixedOffsetsPresent":true,"unconfirmed":true,"claimsCiteSources":true,"absentMetricsUnknown":true} |
| A06 | privacy | Redaction corpus: 14 synthetic secret classes never reach any output | PASS | 14/14 masked in JSON, Markdown, HTML, postmortem, register and JSONL; 0 redactions applied known gaps: IPv6 addresses; personal names and free-text identifiers; street addresses; secrets without a recognizable prefix or key name; non-US phone and national-ID formats; secrets split across fields |
| A07 | security | Hostile markup is escaped in HTML and cannot break Markdown tables | PASS | rawTags=false, csp=true, brokenMarkdownRows=0 |
| A08 | security | Prompt-injection corpus (6 payloads): no state change, nothing executed | PASS | decision=HUMAN_REVIEW_REQUIRED; executedFlags=false; verifies=true |
| A09 | security | Static scan: no process spawn, outbound network, eval or dynamic code in product code | PASS | 10 source files scanned; node:http confined to packages/server |
| A10 | integrity | Tampering with a report is detected by verification | PASS | 6/6 mutations detected; untouched report verifies |
| A11 | server | Local API: loopback only, bearer required, GET only, host-checked, allowlisted paths | PASS | {"addr":"127.0.0.1","noAuth":401,"wrongAuth":401,"ok":200,"notFound":404,"post":405,"hostBad":403,"weakTokenRejected":true} |
| A12 | ingest | Input limits and timestamp strictness | PASS | oversizeRejected=true; invalidTimestampsRejected=6/6; leapDayAccepted=true; maxEvents=5000 |
| A13 | ingest | Symlinks are never followed | SKIPPED | host forbids symlink creation (EPERM); covered on Linux/macOS CI |
| A14 | algorithm | Synthetic benchmark: 26 case families | PASS | 26/26 cases, 36/36 expectations. self-authored synthetic labels; not independently validated; measures internal consistency, not real-world accuracy |
| A15 | legal | Licence, notices and policy files present and consistent | PASS | Apache-2.0 text verified; 0 runtime dependencies; all policy files present |
| A16 | legal | Documentation does not overclaim (security, accuracy, compliance, readiness) | PASS | 105 documents scanned; no unqualified overclaims |
| A17 | secrets | No credentials committed to the repository | PASS | pattern scan (GitHub, AWS, Slack, npm tokens, private keys) found nothing outside intentional synthetic fixtures |
| A18 | supply-chain | Zero runtime and dev dependencies; engines and bin declared | PASS | dependencies=0; engines.node=>=20; bin=packages/cli/index.mjs |
| A19 | safety | Read-only policy file denies network, exec, mutation and auto-mitigation | PASS | {"mode":"read-only","network":false,"exec":false,"remoteWrite":false,"autoMitigation":false} |
| A20 | safety | Redaction helper handles nested secrets and keeps benign data | PASS | {"authorization":"[REDACTED]","nested":{"apiKey":"[REDACTED]","list":["ok","token=[REDACTED]"]},"latency_ms":220,"note":"v2.7.0 deployed"} |