Graincrawl
ReviewAudited by ClawScan on May 17, 2026.
Overview
Graincrawl is a coherent Granola archive helper, but it installs an external CLI and can search or sync sensitive local notes and transcripts.
Install Graincrawl only if you want your agent to access local Granola notes, transcripts, panels, and freshness data. Be cautious with private API syncs, Keychain prompts, and SQL queries, and prefer a pinned or reviewed CLI version if supply-chain assurance matters.
Findings (4)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
The installed binary may change over time when @latest resolves to a newer version.
Installation pulls the latest version of an external Go module, while the reviewed artifacts do not include the CLI source code.
kind: go | module: github.com/openclaw/graincrawl/cmd/graincrawl@latest | creates binaries: graincrawl
Install only from the expected repository, and prefer a pinned or reviewed version if you need stronger supply-chain assurance.
Sensitive Granola content may be brought into the agent conversation or outputs when you ask the skill to search or retrieve it.
The skill is designed to retrieve local Granola notes, transcripts, and panels, which can contain private user or meeting information.
Use local Granola archive data first... graincrawl search "query" ... graincrawl notes --json ... graincrawl transcripts get <id> ... graincrawl panels get <id>
Use it only in workspaces where the agent is allowed to see the relevant Granola notes and transcripts, and review outputs before sharing them.
Syncing or debugging encrypted sources could require access to your Granola account data or local credential prompts.
The documented workflow may involve a private Granola API and Keychain/secrets checks, which can touch account or credential boundaries.
graincrawl sync --source private-api ... Before encrypted source debugging, run explicit unlock/secrets checks; do not surprise-prompt Keychain.
Approve private API syncs and Keychain prompts only when expected, and avoid encrypted-source debugging unless you understand what is being accessed.
SQL queries could reveal more archive information than a simple search if used broadly.
The skill exposes a SQL query path into the archive, but the instructions explicitly limit it to read-only counts and rankings.
graincrawl --json sql "select count(*) as notes from notes;" ... Use read-only SQL for exact counts/rankings.
Keep SQL use limited to read-only, narrowly scoped questions such as counts, rankings, and date ranges.