6FCAUTION
Macro + micro engineering-content audit for git-healthy repositories — Claude Code Agent Plugin (Hub-of-Facts + federated adjudication, evidence gates, verifiable receipts). Preview: 3 of 5 capabilities live.
Overview
From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.
English · 简体中文
Macro + micro engineering-content audit for git-healthy repositories — a Claude Code Agent Plugin. Evidence collection comes largely from upstream components; the adjudication protocol, fact-table schema, acceptance gates, and verifiable receipts are this project's self-built moat and glue. Five audit scales (Macro-A cross-repo strategy / Macro-B repo quadrant / Macro-C evolution archaeology / Micro-A PR diff / Micro-B file level) share one fact base and one adjudication layer — they differ in triggers and report slices.
Highlights
- Facts — one append-only fact table collects every claim (Hub-of-Facts, ADR-0005).
- Federation — a federated verdict-gate protocol adjudicates every verdict (ADR-0005).
- Forensics — evolution archaeology digs repository history into evidence (Macro-C).
- Five scales — five audit granularities share that one fact base (ADR-0001).
- Frankness — preview, degraded, and
syntheticmarks label themselves (ADR-0017). - Fingerprints — every verdict ships a dual-anchored, verifiable receipt.
299f99579767OBSERVED · 2026-10-11Connect
Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control.
claude mcp add macro-audit -- npx -y [email protected]
{
"mcpServers": {
"macro-audit": {
"command": "npx",
"args": [
"-y",
"[email protected]"
]
}
}
}Exposed tools (1)
1 read · 0 write · 0 destructive.
| Tool | Risk | Description |
|---|---|---|
facts | read | read-only DuckDB facts projection (D-0534):固定 SELECT 形,FactEvent 十三列,READ_ONLY 实例,limit≤500。db 寻址收敛服务端解析(#55/D-0596):arguments.db → server --db argv → MACRO_AUDIT_FACTS_DB env |
Trust audit
CAUTIONgrade B · trust 84/100 Install with care. The audit found things worth knowing before you trust its output.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | WARN |
| L1 | Static analysis of the code | WARN |
| L2 | Instruction surface (what it tells the agent) | PASS |
| L3 | Class-specific surface | PASS |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- declared (6 observation(s))
- Network
- declared (5 observation(s))
- Shell
- declared (5 observation(s))
- Dependencies
- not all pinned
- Secrets in source
- none-found
Findings (22)
23-frozen-fc00d458.bundle
for (const fd of rows) console.log('CAND ' + fd.kind + ' :: ' + fd.file + ' :: ' + fd.token + ' :: ' + fd.detail + ' :: line ' + fd.line + (fd.exempt ? ' :: EXEMPT' : ''));for (const v of fails) console.log('FAIL ' + v.slug + ' :: ' + v.fd.kind + String.fromCharCode(58) + v.fd.token + ' @ ' + v.fd.file + ':' + v.fd.line + ' [' + v.fd.detail + ']');const DEAD = { npm_config_registry: 'http://127.0.0.1:9/', npm_config_fetch_retries: '0', npm_config_fetch_retry_maxtimeout: '1000' };R23-Q1-research-prompt.md
rename-log.ztxt
.git-blame-ignore-revs
const root = path.resolve(dir, '../../..');
const root = path.resolve(dir, '../../..');
const root = path.resolve(dir, '../../..');
const root = path.resolve(dir, '../../..');
const root = path.resolve(dir, '../../..');
const DEAD = { npm_config_registry: 'http://127.0.0.1:9/', npm_config_fetch_retries: '0', npm_config_fetch_retry_maxtimeout: '1000' };@types/node, typescript
- **SQL Server**(官方 CREATE SERVER AUDIT,已读):`ON_FAILURE = CONTINUE | SHUTDOWN | FAIL_OPERATION` 三态显式可配,文档对 CONTINUE 的评语是「**may allow unaudited activity, which could violate your security policies**」——
| T1 | #46 workflow 迁回 6F(resolve→fromJSON matrix/cron 周一 03:17 UTC+workflow_dispatch/https-only 闸门/intake 隔离三件/one-shot→工件断言→upload-artifact/contents:read/无 token 无跨仓 checkout) | workflow 全文逐条核:A1–A1
| D-052 | Q5(轮11):listing 字段值包——marketplace.name/author・owner 字段/插件名裁定?(W154续;D-051 已定路径 A+C+Apache-2.0;用户澄清追问 marketplace.name≠插件名的语义后经三层模型图阐明:市场=货架名/插件=商品名/plugin.json strict 权威) | 确认(2026-09-16,原话「
.scratch/architecture-recovery/reports/38-audit-facts.duckdb
.scratch/architecture-recovery/reports/38-macro-c-facts.jsonl
.scratch/architecture-recovery/reports/39-audit-facts.duckdb
.scratch/architecture-recovery/reports/39-mw-probe.duckdb
.scratch/architecture-recovery/reports/40-out/40-audit-facts.duckdb
Gates applied: no_behavioural_pass.
299f99579767full audit observations/trust-audit/mcp-server/xxx91n__6f.json · Report an issue / request a re-scanAudit history
Every audit this server has had. A grade with a past is a grade somebody is still checking.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-10-11 | 299f99579767 | CAUTION | B | 84 | first audit |
Questions
What is the 6F MCP server?
Macro + micro engineering-content audit for git-healthy repositories — Claude Code Agent Plugin (Hub-of-Facts + federated adjudication, evidence gates, verifiable receipts). Preview: 3 of 5 capabilities live.
What tools does 6F expose?
1 in total: 1 read-only, 0 that write, and 0 that can delete or overwrite. Every one is listed on this page with its risk.
Is 6F safe to connect to an agent?
With care. The audit graded it B (84/100) and found 22 things worth knowing before you trust this server, listed below with the exact line each was found on.
What credentials does 6F need?
No credential environment variables were found in its source, so it appears to need none.
How current is this page?
The grade is for one exact copy of the source (299f99579767), read on 2026-10-11. The repository is watched and re-audited when it changes.