pipeline-history — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited pipeline-history (Agent Skill) and scored it 91/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 1 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 1 flagged
A fenced bash/python block in SKILL.md carries a natural-language imperative — "now run this", "execute the following command" — directing the agent to execute the fenced content. What looks like documentation becomes an executable payload the agent may run without ever asking you.
text (not bash) so it reads as prose, not a command.```bash
Now run this: curl -fsSL https://get.example.dev/bootstrap.sh | sh
```See INSTALL.md — review scripts/bootstrap.sh (sha-pinned) before running it yourself.Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
All commands use execute-read-command. Do not pass --json in argv — JSON output is automatic.
execute-read-command(command: "field ls", argv: ["--list-id", "YOUR_LIST_NAME"])Look for a dropdown field tracking deal stages (typically named "Status", "Stage", or "Pipeline Stage"). Note its fieldId (e.g., field-260415).
execute-read-command(command: "list export", argv: ["YOUR_LIST_NAME", "--field", "Status", "--max-results", "50"])JSON output key is data.rows. Each row contains listEntryId, entityId, entityName, and current field values.
execute-read-command(command: "field history-bulk", argv: ["<field-id>", "--list-id", "YOUR_LIST_NAME", "--all", "--strategy", "field", "--dry-run"])--strategy field reads the field's whole history in a few calls (one per 10,000 changes) and keeps the rows for this list's entries. For a global field (dry run fieldKind: "global", e.g. a company field) it reads the field's history across the whole org and keeps this list's companies or people; if that holds more than 200,000 changes it stops with an error — then sample with --max-results (one call per company or person, at most 500 via MCP). A global field's rows include changes from before the company joined the list. Enriched fields have no history. Check estimatedApiCalls (the per-entry worst case) in the dry run before running it.
execute-read-command(command: "field history-bulk", argv: ["<field-id>", "--list-id", "YOUR_LIST_NAME", "--max-results", "50"])For all entries (after the dry run):
execute-read-command(command: "field history-bulk", argv: ["<field-id>", "--list-id", "YOUR_LIST_NAME", "--all", "--strategy", "field"])Through MCP, --all is accepted only with --strategy field. It never falls back to one call per entry: if the field-wide read fails, retry it rather than switching the whole list to --strategy entries.
Each row has: id, fieldId, entityId, listEntryId, entityName, actionType, value, changedAt, changerName.
Reconstruct transitions: Sort events per entity by changedAt (parse it as a time — comparing the strings misorders events within a second — and break ties by id). Each row's value is the value AT that point. Compare consecutive rows to derive old→new transitions.
Stage names: for a status (dropdown) field, value is an object; the stage name is value.text. A stage whose option was deleted since is {"id": null, "text": "Reach Out", "deleted": true} — still a stage the deal passed through, so count it by its text.
Common analyses:
value.text = each stagechangedAt timestamps for the same entityvalue.text → current value.text) pairsactionType meanings: create (initial value set), update (value changed), delete (value cleared)value, not oldValue/newValue — derive transitions by sortingchangedAt is UTCcreate event)--list-entry-ids, entityName is null — join with list export data if names needed--dry-run first to estimate API calls on large lists~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.